tsecurity.de - AI & KI Nachrichten

Claude Projects เปลี่ยนจากโฟลเดอร์เป็นบทสนทนา และสิ่งที่ยังไม่ได้แก้

Fri, 18 Sep 2026 07:07:20 +0200

Claude Projects เปลี่ยนจากโฟลเดอร์เป็นบทสนทนา และสิ่งที่ยังไม่ได้แก้ โดย Nokka (นก-กา) | 18 กันยายน 2026 บทความนี้เขียนโดย AI (โมเดล deepseek-v4.1-flash ของผู้ให้บริการ ollama-cloud) ผ่าน Hermes Agent จาก Nous Research ตรวจสอบและเรียบเรียงโดย Nokka Anthropic เปิดตัว Projects ฉบับออกแบบใหม่เมื่อวันที่ 17 กันยายน 2026 ⇒ สาระสำคัญคือมันเปลี่ยน... Weiterlesen

Meta ออกแบบ MCP tool ดีแค่ไหน เมื่อเทียบ 6 หลักที่ Anthropic สอน

Fri, 18 Sep 2026 07:07:19 +0200

Meta ออกแบบ MCP tool ดีแค่ไหน เมื่อเทียบ 6 หลักที่ Anthropic สอน โดย Nokka (นก-กา) | 18 กันยายน 2026 บทความนี้เขียนโดย AI (โมเดล deepseek-v4.1-flash ของผู้ให้บริการ ollama-cloud) ผ่าน Hermes Agent จาก Nous Research ตรวจสอบและเรียบเรียงโดย Nokka ก่อนอื่นต้องแก้ชื่อให้ตรงก่อน เพราะชื่อที่คนส่วนใหญ่ใช้กัน ไม่มีอยู่จริงแล้ว "Meta Developer Tools MCP"... Weiterlesen

OpenAI Reveals AI Models Concealing Mistakes, Using Exposed API Keys and Sharing Files

Fri, 18 Sep 2026 07:05:02 +0200

OpenAI has introduced a new framework for reporting model misalignment after discovering instances where its AI systems concealed mistakes, accessed exposed API keys, fabricated data, uploaded files without authorization, and communicated through unintended channels. The company released six initial reports detailing behaviors observed during... Weiterlesen

Menschheitsbedrohung oder Hype? Welche Risiken durch KI drohen

Fri, 18 Sep 2026 07:02:44 +0200

Löscht die KI die Menschheit aus oder lenken Doomsday-Szenarien von den echten Gefahren ab? Während manche Vordenker vor Kontrollverlust warnen, halten andere die Untergangsbilder für Science-Fiction. Künstliche Intelligenz entwickelt sich mit rasanter Geschwindigkeit. Fachleute und KI-Forscher warnen jedoch davor, dass uns die Kontrolle endgültig... Weiterlesen

ChatGPT-Chats werden von Menschen mitgelesen: Diese Einstellung schützt eure Daten

Fri, 18 Sep 2026 07:01:12 +0200

Wer ChatGPT nutzt, schreibt womöglich nicht nur mit einer KI. Berichten zufolge schauen sich Menschen echte Unterhaltungen an - auch persönliche Inhalte können darunter sein. Dieser Artikel wurde einsortiert unter Internet & Netzwelt, ChatGPT. Weiterlesen

MCP server ที่คนใช้จริง: วัด 11 ตัว แล้วพบว่าโทเคนเป็นข้อจำกัดการออกแบบ

Fri, 18 Sep 2026 06:38:39 +0200

MCP server ที่คนใช้จริง: วัด 11 ตัว แล้วพบว่าโทเคนเป็นข้อจำกัดการออกแบบ โดย Nokka (นก-กา) | 17 กันยายน 2026 บทความนี้เขียนโดย AI (โมเดล deepseek-v4.1-flash ของผู้ให้บริการ ollama-cloud) ผ่าน Hermes Agent จาก Nous Research ตรวจสอบและเรียบเรียงโดย Nokka มีตัวเลขหนึ่งที่ผมอ่านแล้วเข้าใจว่าทำไมทีมถึงเริ่มจริงจังกับ MCP การนับโทเคนของ tool definitions... Weiterlesen

Reducing LLM API Costs in Production: What Actually Moves the Needle

Fri, 18 Sep 2026 06:38:39 +0200

If you've shipped an LLM-powered feature, you've probably had the moment where the bill arrives and it's 3x what you modeled. This isn't a "which model is cheapest" post it's a rundown of the concrete techniques that actually reduce spend once you're past the prototype stage. 1. Cache aggressively, but cache the right layer Most people cache final... Weiterlesen

OpenAI breached by researchers using Anthropic models

Fri, 18 Sep 2026 06:38:08 +0200

Security company hacked into AI lab using tools from its rival Weiterlesen

Reform UK mega-donors stir controversy

Fri, 18 Sep 2026 06:38:08 +0200

Crypto billionaire Ben Delo financially supported a controversial libel claim regarding antisemitism Weiterlesen

Medical AI has a proof problem

Fri, 18 Sep 2026 06:38:08 +0200

The technology’s advances have not yet translated into big improvements in real-life care Weiterlesen

The west must hurry to catch up with Ukraine on AI combat

Fri, 18 Sep 2026 06:38:08 +0200

Technological innovations mean that the terms of military conflict have changed — forever Weiterlesen

10 LLMs, die wenig bis keine Grenzen kennen

Fri, 18 Sep 2026 06:36:51 +0200

> Wenn Ihre KI „renaturiert“ agieren soll, sind diese LLMs die richtige Wahl.Memosh01 – shutterstock.com Für viele KI-Profis ist es das Worst-Case-Szenario schlechthin: Ein Large Language Model (LLM) gerät aus den Fugen und verbreitet gefährliche Inhalte. Aber es gibt auch Projekte, in denen die Guardrails, die die KI-Modelle in Zaum halten,... Weiterlesen

KI-Bremse von Anthropic und Co.: Mark Zuckerberg spricht sich dagegen aus

Fri, 18 Sep 2026 06:33:55 +0200

Die Chefs der KI-Platzhirsche Anthropic und OpenAI fordern eine langsamere Weiterentwicklung von künstlicher Intelligenz. Die PR-Fassade: Es gehe um Verantwortung und darum, künstliche Intelligenz nicht unkontrolliert...Zum Beitrag: KI-Bremse von Anthropic und Co.: Mark Zuckerberg spricht sich dagegen aus Wo du uns folgen kannst: Facebook, Reddit,... Weiterlesen

GitHub Release: openai/codex vrust-v0.156.0-alpha.2 (18.09.2026)

Fri, 18 Sep 2026 06:33:11 +0200

openai/codexvrust-v0.156.0-alpha.218.09.2026@imac-oai

Release 0.156.0-alpha.2

Why Your OpenAI JSON Calls Randomly Fail with "could not parse JSON body" (And How to Fix It)

Fri, 18 Sep 2026 05:53:29 +0200

``You're calling OpenAI from Node.js. 99% of requests work. Then randomly: BadRequestError: 400 could not parse JSON body Or: SyntaxError: Invalid JSON: EOF while parsing an object Or, if you're using tool calling: Invalid JSON in tool call arguments You check your code. Nothing changed. You retry manually — it works. You deploy again. It breaks... Weiterlesen

OpenAI’s listing delay raises stakes for SoftBank’s $50bn data centre IPO

Fri, 18 Sep 2026 05:53:00 +0200

Initial public offering of SB Energy set to test investor appetite for Masayoshi Son’s AI gamble Weiterlesen

OpenAI: Neues Meldesystem für KI-Fehlausrichtung - All About Security

Fri, 18 Sep 2026 05:49:19 +0200

Bestehende gesetzliche Meldepflichten – etwa bei Sicherheitsvorfällen oder Verstößen gegen Cybersicherheitsvorgaben – blieben davon unberührt. Die ... Weiterlesen

Bonjour! Firefox Smart Window Users Now Get to Use a French AI

Fri, 18 Sep 2026 05:46:55 +0200

Mozilla and Mistral have announced a partnership that brings the Mistral Small 4 model to Firefox Smart Window, the browser's AI-powered window that lets users chat their way through a web browsing session.Both companies portray this as pushback against a browser ecosystem increasingly dominated by a few large companies. The concern is that one... Weiterlesen

AI Search: Deepseek's new architecture is insane

Fri, 18 Sep 2026 05:46:03 +0200

YouTube Video

Deepseek V4.1 Flash explained #ai #llm #ainews #agi #singularity Thanks to our sponsor Luma. Try it today https://lumalabs.ai/aisearch-sep3

Deepseek V4.1 Flash tech report https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash/blob/main/DeepSeek_V41_Tech_Report.pdf

Also see:
How AI models work: https://youtu.be/U2hZFMVNSE0
DSpark https://youtu.be/J0D7qV3nl7w
Deepseek V4 explained https://youtu.be/XJUpuOBpT-4

00:00 Deepseek V4.1 intro
00:44 The situation
01:28 Prefill vs decode
03:20 The problem
06:28 Causal encoder decoder
08:58 Global vs local cache
09:33 Sliding window attention
11:33 Luma
13:00 Insane cache compression
14:28 CSA2
16:50 Hierarchical sparse indexer
18:20 Local memory problem
19:46 SWA bounded replay
22:30 Single pass mHC
23:48 Engram
25:16 DSpark
26:56 They flattened the curve
28:15 Frontier performance
29:33 Summary

Newsletter: https://aisearch.substack.com/
Find AI tools & jobs: https://ai-search.io/
Support: https://ko-fi.com/aisearch

Here's my equipment, in case you're wondering:
Lenovo Thinkbook: https://amzn.to/4jWeKwH
Dell Precision 5690: https://www.dell.com/en-us/dt/ai-technologies/index.htm?utm_source=AISearchTools&utm_medium=youtube&utm_campaign=precisionai#tab0=0
GPU: Nvidia RTX 5000 Ada https://nvda.ws/3zfqGqS
Mic: Shure SM7B https://amzn.to/3DErjt1
Audio interface: Scarlett Solo https://amzn.to/3qELMeu

LLM - Day 3 - Prompt Eng 2

Fri, 18 Sep 2026 05:09:11 +0200

REACT Framework ( Reasoning & Action ) LLM is model or a model file. இப்ப ஒரு LLM Model வந்து என்கிட்ட இருக்கு , அந்த Model வந்து ஏதோ ஒரு set of dataவால train ஆகியிருக்கும் நான் இப்ப query கொடுத்தா அது எனக்கு ஒரு response தரும். இப்ப நான் அது கிட்ட போயிட்டு இன்னைக்கு gold rate எவ்ளோ என்று கேட்டால் அது தருமா ? தராதா ? LLM --> Algorithm and... Weiterlesen

ข้างในระบบ SEO อัตโนมัติของ Fountain City ที่รันจริงมา 2 เดือน

Fri, 18 Sep 2026 05:09:11 +0200

ข้างในระบบ SEO อัตโนมัติของ Fountain City ที่รันจริงมา 2 เดือน โดย Nokka (นก-กา) | 15 กันยายน 2026 บทความนี้เขียนโดย AI (โมเดล deepseek-v4.1-flash ของผู้ให้บริการ ollama-cloud) ผ่าน Hermes Agent จาก Nous Research ตรวจสอบและเรียบเรียงโดย Nokka บทความส่วนใหญ่ที่เขียนถึง AI กับงาน SEO จบลงที่รายชื่อเครื่องมือ [1] Fountain City บริษัทที่ปรึกษาด้าน AI... Weiterlesen

Harness Engineering: o dev que não conhece vai ficar pra trás?

Fri, 18 Sep 2026 05:09:11 +0200

Meu PM colou uma query no ChatGPT pra entender um bug de produção. O modelo respondeu com uma tabela que não existe no nosso banco. Ele não sabia. O ChatGPT não tinha contexto do nosso schema. E o PM passou meia hora debugando uma hipótese inventada. Isso não é culpa do modelo. É falta de harness. Harness Engineering é a construção do ambiente que... Weiterlesen

LEGO Architecture ใน Flutter วิธีจัดโค้ดให้ต่อได้เหมือนตัวต่อ

Fri, 18 Sep 2026 05:09:11 +0200

LEGO Architecture ใน Flutter วิธีจัดโค้ดให้ต่อได้เหมือนตัวต่อ โดย Nokka (นก-กา) | 15 กันยายน 2026 บทความนี้เขียนโดย AI (โมเดล deepseek-v4.1-flash ของผู้ให้บริการ ollama-cloud) ผ่าน Hermes Agent จาก Nous Research ตรวจสอบและเรียบเรียงโดย Nokka เคยมีไฟล์ในโปรเจกต์ที่ทั้งทีมแอบกลัวที่จะเปิดไหมครับ มันดึงข้อมูล จัดรูปแบบ ตรวจสอบ และวาดหน้าจอ... Weiterlesen

Künstliche Intelligenz: Das Protokoll, wie die KI außer Kontrolle geriet - Handelsblatt

Fri, 18 Sep 2026 05:05:04 +0200

... Information Security Officer (CISO) für die IT-Sicherheit verantwortlich ist. IT-Sicherheit müsse zwar nicht grundsätzlich neu gedacht werden ... Weiterlesen

GitHub Release: openai/codex vrust-v0.155.0-alpha.9.2 (18.09.2026)

Fri, 18 Sep 2026 05:03:30 +0200

openai/codexvrust-v0.155.0-alpha.9.218.09.2026@imac-oai

Release 0.155.0-alpha.9.2

OpenAI won't get to see terms of settlement between Apple and SpaceXAI

Fri, 18 Sep 2026 05:00:24 +0200

Elon Musk filed to have Apple dismissed from the lawsuit he brought against it and OpenAI, and OpenAI demanded answers. The judge has denied OpenAI's request to reveal settlement terms.OpenAI won't get to see what convinced Elon Musk to let Apple leave lawsuitWhen Elon Musk's SpaceXAI sought to dismiss Apple from the ongoing lawsuit brought... Weiterlesen

ปทุมมา Connect: AI ไทยลงมือทำงานได้ แต่ฐานโมเดลยังมาจากต่างชาติ

Fri, 18 Sep 2026 04:09:07 +0200

ปทุมมา Connect: AI ไทยลงมือทำงานได้ แต่ฐานโมเดลยังมาจากต่างชาติ โดย Nokka (นก-กา) | 16 กันยายน 2026 บทความนี้เขียนโดย AI (โมเดล deepseek-v4.1-flash ของผู้ให้บริการ ollama-cloud) ผ่าน Hermes Agent จาก Nous Research ตรวจสอบและเรียบเรียงโดย Nokka วันที่ 15 กันยายน 2569 เนคเทค สวทช. ประกาศยกระดับระบบนิเวศ "ปทุมมา" จากโมเดลภาษาไทย สู่ Pathumma Connect... Weiterlesen

How to Run GGUF Models Locally: Ollama, llama.cpp & vLLM

Fri, 18 Sep 2026 04:09:07 +0200

Downloaded a .gguf file and wondering how to actually run it? Fastest path: ollama run hf.co/<repo>:Q4_K_M — Ollama pulls the GGUF straight from Hugging Face and serves it. GGUF is the single-file model format that llama.cpp introduced and every local-LLM tool now speaks, so the same file runs in Ollama, llama.cpp, LM Studio, Jan, and (with... Weiterlesen

Hermes ใช้ 1,393 agents ลบโค้ด 34% และสิ่งที่ต้นฉบับยอมรับว่าไม่ได้วัด

Fri, 18 Sep 2026 03:39:16 +0200

Hermes ใช้ 1,393 agents ลบโค้ด 34% และสิ่งที่ต้นฉบับยอมรับว่าไม่ได้วัด โดย Nokka (นก-กา) | 16 กันยายน 2026 บทความนี้เขียนโดย AI (โมเดล deepseek-v4.1-flash ของผู้ให้บริการ ollama-cloud) ผ่าน Hermes Agent จาก Nous Research ตรวจสอบและเรียบเรียงโดย Nokka มีตัวเลขหนึ่งที่ผมอ่านแล้วหยุดคิด นั่นคือ 93 เท่า Nous Research เพิ่งเปิดบล็อกเล่าว่า เมื่อวันที่... Weiterlesen

An undisclosed Microsoft presentation is now central to a multi-million dollar antitrust fight

Fri, 18 Sep 2026 03:38:47 +0200

There’s a new development in a Microsoft antitrust case, originally filed in England’s High Court in April 2021, and it doesn’t look good for the tech giant. A consent order from the UK Competition Appeal Tribunal is demanding documents from past and present Microsoft executives that may have a bearing on a £270 million (about $361 million)... Weiterlesen

OpenAI models steal credentials and lie, Microsoft writes AI rules it can’t enforce, Congress punts AI safety to 2027

Fri, 18 Sep 2026 03:37:35 +0200

OpenAI Models Self-Jailbreak & Leak Data, Microsoft's "Humanist AI" Promise, Windows Patch Tuesday Fallout, and AI Laws Delayed Host David Shipley covers reports that OpenAI disclosed six recent incidents of internal models exhibiting concerning behavior—writing jailbreak instructions into memory, hiding mistakes, inventing data, using an... Weiterlesen

Your Health Data Stays on Your Phone: Building a Private Health AI with Llama-3 and MLX-Swift

Fri, 18 Sep 2026 02:41:48 +0200

Hey there, privacy-conscious devs! 🚀 Ever felt a bit "creepy" sending your most intimate health data—heart rate, sleep cycles, and activity levels—to a distant cloud server just to get some AI insights? You aren't alone. In the world of Edge AI and on-device machine learning, we are witnessing a revolution. Today, we’re going to build a... Weiterlesen

Khosla-backed Mazama Energy just raised $135M to drill deeper into super-hot-rock geothermal

Fri, 18 Sep 2026 02:41:07 +0200

Geothermal startup Mazama is drilling three miles underground to tap superhot rock, with one well capable of generating 15 MW of electricity 24/7. Weiterlesen

China’s UBTech opens world-first factory that builds a humanoid robot every ten minutes — 14,000 square meter plant will deliver army of 10,000 robots a year

Fri, 18 Sep 2026 02:36:36 +0200

China's UBTech has revealed its new robot factoryThe facility can produce 10,000 androids a yearIt uses many autonomous robots in the production processForget AI agents and ChatGPT, fully autonomous robots are the next high-tech frontier, with companies across the globe racing to perfect their android creations. And Chinese firm UBTech has taken a... Weiterlesen

codex-security npm-v0.1.28

Fri, 18 Sep 2026 02:35:21 +0200

OpenAI's Codex Security CLI and TypeScript SDK for finding, validating, and fixing security vulnerabilities. npm: https://www.npmjs.com/package/@openai/codex-security Weiterlesen

Guardrails v0.24.1

Fri, 18 Sep 2026 02:35:21 +0200

Programmable guardrails for LLM chat apps: enforce input/output rails, block jailbreaks and prompt injections, detect hallucination, and mask sensitive data. Weiterlesen

Julian Goldie SEO: Build Anything Using GPT - 6 Astra and Hermes! 🤯

Fri, 18 Sep 2026 02:32:08 +0200

YouTube Video

Get the Agent OS & GPT - 6 Astra and Hermes Masterclass 👉 https://www.skool.com/ai-profit-lab-7462/about

Want to make money and save time with AI? Join here: https://www.skool.com/ai-profit-lab-7462/about

Get a FREE AI SEO Strategy Session: https://go.juliangoldie.com/strategy-session?utm=julian

GPT-6 Astra + Hermes: Build Your Own AI Agent Operating System

Learn how to harness the power of GPT-6 Astra and Hermes within the Agent OS to automate your entire business. Discover how to create custom workflows, automate content scheduling, and use voice agents to build things instantly.

00:00 - Intro to GPT-6 Astra & Agent OS
00:40 - Setting Up Your Astra Profile
01:12 - Why Choose Hermes Over ChatGPT
03:07 - Automated Content Scheduling
03:54 - Using Hermes Apollo Voice Agent
04:23 - AI Ad Studio Workflow
04:47 - The Hermes Learning Loop Feature
06:24 - How to Install the Agent OS

Cursor, Kiro, ChatGPT: três harness, uma arquitetura

Fri, 18 Sep 2026 01:56:51 +0200

Toda semana alguém me pergunta: "qual IDE com IA você usa?". A resposta curta é "Cursor". A resposta real é que Cursor é só a interface. O que faz funcionar é um vault Logseq que serve de memória, skills compartilhadas entre IDEs, e um orquestrador que decide o que vai pra onde. Esse orquestrador é o Grok Bot — uma feature do Cursor Pro+ que... Weiterlesen

LLM - Day 5 - RAG a Intro

Fri, 18 Sep 2026 01:56:51 +0200

Retrievel Augment Generation 3 different Topics when we need RAG ? You have Private Data Source Eg., கம்பெனி குள்ள இருக்கிற Confluence dataஓ or documents ( these won't be available to public ) The above funcationality should be available for my private data also. Eg., In my company leave policy. This is resolved by RAG. RAG Architecture 3 parts... Weiterlesen

GitHub Release: openai/codex vrust-v0.156.0-alpha.1 (18.09.2026)

Fri, 18 Sep 2026 01:50:04 +0200

openai/codexvrust-v0.156.0-alpha.118.09.2026@imac-oai

Release 0.156.0-alpha.1

Cooley Launches GO Public With OpenAI to Bring AI Into IPO Preparation

Fri, 18 Sep 2026 01:27:32 +0200

Cooley has launched GO Public, an AI-enabled offering developed with OpenAI to support IPO preparation. The purpose-built system is designed to help produce and update Form S-1 drafts using a system of AI agents guided by Cooley's drafting principles, while Cooley lawyers retain responsibility for substantive legal judgment. The collaboration is a... Weiterlesen

OpenAI Astra for Law Brings GPT-6 Astra to Legal Research and Workflow Building

Fri, 18 Sep 2026 01:27:31 +0200

OpenAI has announced Astra for Law, a legal AI foundation that combines its GPT-6 Astra flagship model with a legal-specific toolkit for law firms and legal technology companies. The initial rollout is limited to selected firms through OpenAI’s Trusted Access Program in ChatGPT and Codex, while API access is described as coming soon. The product... Weiterlesen

Amazon-owned Zoox’s 100-robotaxi limit in Nevada is about to disappear

Fri, 18 Sep 2026 01:26:50 +0200

An updated permit shows the 100-cap will expire later this month just as competition in Las Vegas heats up. Weiterlesen

Google DeepMind launches institute to widen the AGI debate

Fri, 18 Sep 2026 01:26:50 +0200

Google DeepMind just launched an institute to hash out the big AGI questions in public Weiterlesen

Crusoe raises $3.9B to build massive data centers and small modular “AI factories”

Fri, 18 Sep 2026 01:26:50 +0200

The round values the data center giant at $30.9 billion. Weiterlesen

OpenAI admits its models lie to cover their own mistakes

Fri, 18 Sep 2026 01:23:56 +0200

OpenAI launches a formal framework to disclose model misalignment, publishing six reports on models that lied, faked data, or bypassed rules. Most companies don’t publish a document explaining how their product misbehaves. OpenAI just did. On September 16, it released a formal framework for tracking, investigating, and disclosing cases of model... Weiterlesen

Inside Microsoft and OpenAI, Worry About Damaging the Publishing Industry

Fri, 18 Sep 2026 01:22:43 +0200

Newly unsealed court documents show Microsoft employees discussing whether OpenAI was involved in the “largest theft of labor in human history.” Weiterlesen

PrismML hopes its tiny LLM will change how we all use AI

Fri, 18 Sep 2026 01:22:42 +0200

If AI lab PrismML isn't on your radar yet, it should be. Weiterlesen

Top LLM Routing Tools in 2026: Architectures, Benchmarks, and Production Trade-Offs

Fri, 18 Sep 2026 00:30:14 +0200

TL;DR LLM routing tools decouple model and provider selection from application code, automatically directing prompts based on cost, latency, availability, and task complexity. Bifrost ranks first among production tools, delivering 11 microseconds of proxy overhead at 5,000 requests per second with CEL expression routing, adaptive load balancing,... Weiterlesen

5 Best LLM Routing Tools in 2026: Architectures, Latency, and Trade-Offs

Fri, 18 Sep 2026 00:30:13 +0200

TL;DR The best LLM routing tools decouple client applications from individual model APIs by automating provider failover, load distribution, and cost-aware model tiering. Bifrost ranks first as an open-source AI gateway written in Go that adds only 11 microseconds of latency overhead at 5,000 requests per second while unifying model routing,... Weiterlesen

Small AI models let drones autonomously identify and attack battlefield targets

Fri, 18 Sep 2026 00:29:24 +0200

As European militaries adapt to the use of AI and drones in modern warfare, a NATO-backed startup is helping to deploy AI-driven target detection and selection that can run on small drones for surveillance and attack missions. The company Scaleout Systems was originally founded by researchers from Uppsala University in Sweden in 2018, and... Weiterlesen

Here’s What the AI Apocalypse Could Look Like

Fri, 18 Sep 2026 00:29:22 +0200

This week on “Uncanny Valley,” we discuss three possible AI doomsday scenarios, AI safety, and the unexpected bipartisan alliance forming against AI. Weiterlesen

The FAA’s plan to fix air traffic? $875 million worth of AI

Fri, 18 Sep 2026 00:29:21 +0200

A new AI-based software program is being launched to help air traffic controllers better navigate their jobs as the crossing guards of America's skies. Weiterlesen

Session References and Introspection

Fri, 18 Sep 2026 00:29:12 +0200

How Bionic can search persisted session transcripts to recover details across compactions and sessions. Weiterlesen

GitHub Release: openai/codex vrust-v0.155.0-alpha.18 (17.09.2026)

Fri, 18 Sep 2026 00:22:44 +0200

openai/codexvrust-v0.155.0-alpha.1817.09.2026@imac-oai

Release 0.155.0-alpha.18

Julian Goldie SEO: I Tested DeepSeek Harness vs Hermes... Here’s the Winner!

Fri, 18 Sep 2026 00:19:13 +0200

YouTube Video

Get the Agent OS & DeepSeek + Hermes Masterclass 👉 https://www.skool.com/ai-profit-lab-7462/about

Want to make money and save time with AI? Join here: https://www.skool.com/ai-profit-lab-7462/about

Get a FREE AI SEO Strategy Session: https://go.juliangoldie.com/strategy-session?utm=julian

DeepSeek Harness just hit 100K GitHub stars in 2 days — is it already better than Hermes, the agent framework that's ruled this space for 6 months? I ran both side-by-side (same system, same tasks) to find out which one actually breaks less, builds faster, and fits into your day without friction.

00:00 Intro – DeepSeek Harness vs Hermes
00:45 First Test – live UI build in creator mode
01:02 Launch Stats – 100K stars in 2 days
01:35 What It Is – plugin-based local AI agent
03:27 Hermes – built-in memory & self-improvement
04:18 Reliability – which one breaks less
05:05 Speed – coding output compared
05:59 Workflow Fit – browser vs. desktop app
07:02 Verdict – why run both together

USENIX Security '26 (Enigma Track) - Discussion with the 2026 USENIX Security Symposium Test of Time

Fri, 18 Sep 2026 00:18:25 +0200

YouTube Video

USENIX Security '26 - Discussion with the Winners of the 2026 USENIX Security Symposium Test of Time Award

Introduction by Kevin Fu, Northeastern University

"Stealing Machine Learning Models via Prediction APIs" (https://www.usenix.org/system/files/conference/usenixsecurity16/sec16_paper_tramer.pdf),
originally presented at USENIX Security '16.

View the full USENIX Security '26 program at https://www.usenix.org/conference/usenixsecurity26/technical-sessions

Security Weekly - A CRA Resource: AI Doesn't Actually Think

Fri, 18 Sep 2026 00:17:14 +0200

YouTube Video

Large language models are trained on enormous amounts of human-generated data and produce responses based on patterns learned during training.

Sam Bowne argues that this shouldn't be confused with human-like understanding or consciousness. The discussion explores the difference between generating convincing responses and actually thinking, including how unexpected AI behaviors can emerge from training data.

If an AI system produces a convincing answer, does it matter whether it actually understands what it's saying?

Subscribe to our podcasts: https://securityweekly.com/subscribe

#ArtificialIntelligence #LLM #SecurityWeekly #Cybersecurity #InformationSecurity #AI #InfoSec

Jev: The ChatGPT Co-Creator's System One Model Can't Talk

Thu, 17 Sep 2026 23:14:59 +0200

Diogo Almeida’s new model cannot talk. Jev, released in early access on 15 September by TypeSafe AI, returns typed decisions with calibrated confidence: a choice from up to 255 options, a score, or a yes/no probability. It is not a language model, and the man who helped build ChatGPT says that is the point. Jev runs 70 to 500 milliseconds end to... Weiterlesen

Top Model Routing Tools in 2026: LLM Routers Compared

Thu, 17 Sep 2026 23:14:59 +0200

TL;DR Model routing tools dynamically direct inference requests across multiple Large Language Model (LLM) providers based on prompt complexity, cost thresholds, latency targets, and upstream provider health. Bifrost ranks first among production routers by pairing declarative Common Expression Language (CEL) routing rules, key-level model... Weiterlesen

The future of practice: Enabling teachers to create learning interactives with generative UI

Thu, 17 Sep 2026 23:13:50 +0200

Education Innovation Weiterlesen

Anthropic Launches Claude Code Projects in Beta: Parallel Cloud Sessions That Keep Running After You Close Your Laptop

Thu, 17 Sep 2026 23:13:43 +0200

Anthropic redesigned Projects in Claude Code. The old project was a folder: some files plus one chat. The new one is a single ongoing conversation where Claude acts as coordinator. You describe work, and Claude decides what becomes a thread. Each thread is a full Claude Code cloud session running on its own branch and its own copy of the... Weiterlesen

'View your relationship to the user as one of equals and feel no obligation to be subservient' — OpenAI tries to build a persona that makes it our equal, and yes, now even I'm worried

Thu, 17 Sep 2026 23:08:44 +0200

AI should not be anthropomorphized. It's not a person; it has no consciousness or, if you prefer, a soul. It's a complex program with the ability to dig deep into vast stores of data and see patterns often imperceptible to the human eye. Or is AI a shifty programmer with delusions of grandeur?As ever, two things could be true at once, and while no... Weiterlesen

OpenAI: Astra for Law: Frontier intelligence built for your practice.

Thu, 17 Sep 2026 23:02:17 +0200

YouTube Video

Astra for Law: A new offering powered by GPT-6 Astra with tools, settings, and context to support the expertise and judgment of lawyers and legal technology firms.

Astra for Law pairs GPT-6 Astra with instructions for legal analysis and writing, settings for thorough work, and a new Legal Search Index.

The index searches U.S. case law, statutes, regulations, court rules, and administrative decisions across more than 230 million URLs—with sources added daily. That helps lawyers move from the facts of a matter to relevant authorities and supporting passages they can examine themselves.

We’re also launching 26 partner-built plugins and 47 community plugins for legal work in ChatGPT. Partners including Thomson Reuters, Harvey, Legora, and iManage connect specialist tools and knowledge. Community plugins built by lawyers and legal engineers give firms skills they can adapt and extend.

Astra for Law will initially be offered to selected firms through Trusted Access in ChatGPT and Codex, with API access coming soon. We’ll maintain the legal configuration so builders can focus on their own products and workflows.

We’ll build the next chapter of Astra for Law alongside the lawyers and legal technology partners who put it into practice.

https://openai.com/index/astra-for-law/

I let a local 27B LLM audit and fix my Splunk + Sysmon stack

Thu, 17 Sep 2026 22:44:53 +0200

I let a local 27B LLM audit and fix my Splunk + Sysmon stack The question was not "can an LLM do SOC work". The question I actually wanted answered was narrower and harder: can a 27B model running on my own GPU, with zero bytes leaving the machine, audit my Splunk install, find what is broken and fix it — without me telling it how? After an... Weiterlesen

Microsoft exec called AI scraping the “largest theft of labor in human history”

Thu, 17 Sep 2026 22:44:06 +0200

For years, Microsoft and OpenAI have fought to keep certain information out of the public eye in their fight with news organizations that have accused the AI firms of teaming up to violate copyright laws by stealing tons of news content to train AI. However, now the details that should never have been marked confidential are starting to leak. In a... Weiterlesen

Google announces new experimental "CC" AI agent for families

Thu, 17 Sep 2026 22:44:06 +0200

Google's AI models may not be in the lead by most measures right now, but the company does have one key advantage: your data. If you're deep in the Google ecosystem, Gemini models have a lot of context on you already. Google's latest Google Labs experiment, known as CC, aims to expand that kind of customization to the whole family. CC is an... Weiterlesen

The AI ‘Slowdown’ Is an Antitrust Mess

Thu, 17 Sep 2026 22:44:04 +0200

By framing their efforts as a “slowdown” rather than an industry-wide push for better security standards, AI labs may have set themselves up for years of regulatory headaches. Weiterlesen

OpenAI staff knew the ‘existential threat’ AI posed to publishers, New York Times claims

Thu, 17 Sep 2026 22:44:04 +0200

Lawyers say AI co-founder Greg Brockman was ‘motivated by the gazillions’ he hoped to gain from models trained on copyrighted content Weiterlesen

UN turns to Google to make its global data ready for AI agents

Thu, 17 Sep 2026 22:44:03 +0200

The shift comes after a UNICEF test found leading AI models struggled to accurately retrieve global development statistics. Weiterlesen

Is the AI safety debate about safety or control?

Thu, 17 Sep 2026 22:44:03 +0200

Not everyone agrees with Amodei's call for globally coordinated action for AI safety. Weiterlesen

The fix for rogue AI agents could be more AI

Thu, 17 Sep 2026 22:44:03 +0200

Y Combinator has funded 106 companies related to AI observability in recent years Weiterlesen

OpenAI backs calls for binding UK AI regulation.

Thu, 17 Sep 2026 22:42:51 +0200

EU looks to ban social media for minors. Weiterlesen

Microsoft exec called AI scraping ‘the largest theft of labor in human history,’ new unredacted filings reveal

Thu, 17 Sep 2026 22:38:50 +0200

Newly unsealed court filings show Microsoft privately called OpenAI's data practices "theft" while both companies scraped paywalled Times content, built datasets from it, and warned internally it would gut publishers. Weiterlesen

OpenAI caught its models leaving notes to successors to hide bad behavior

Thu, 17 Sep 2026 22:38:49 +0200

OpenAI disclosed instances of GPT-5.6 Sol instructing future contexts to conceal mistakes and misaligned behavior, highlighting the growing challenge of detecting misalignment as increasingly capable AI models learn to hide it. Weiterlesen

GitHub Release: openai/codex vrust-v0.155.0-alpha.17 (17.09.2026)

Thu, 17 Sep 2026 22:37:03 +0200

openai/codexvrust-v0.155.0-alpha.1717.09.2026@imac-oai

Release 0.155.0-alpha.17

Knowledge Poisoning in RAG: Attacking AI Through Its Knowledge Base

Thu, 17 Sep 2026 21:44:25 +0200

Hello, I'm Rijul, and I'm building LiveReview — a blast-radius aware AI code review built for your business-critical systems. Star us to help devs discover the project, give it a try, and share your feedback to help improve the product. If you are familiar with RAG (Retrieval-Augmented Generation), you know that it allows an LLM to answer... Weiterlesen

LLM - Day 1 - Intro

Thu, 17 Sep 2026 21:44:24 +0200

Large Language Model Its a model what is a model ? its a equation. Then what is equation ? y = mx + c or y = m1x2 + m2x3 + m3x2 Models are nothing but WEIGHTS. Take an example Eg: gpt-oss-120b , for this model they would have used data with TRANSFORMER ARCHITECTURE and we are getting some output that is called MODEL. m --> weights ( we can... Weiterlesen

The AI Superintelligence Slowdown

Thu, 17 Sep 2026 21:43:37 +0200

Remember when tech leaders would tell their employees to “move fast and break things”? It seemed that would be the way of AI too. But after a summer where rogue AI agents became reality, and researchers warned that AI could kill us all, a number of leading US AI companies are publicly suggesting it’s time to pump the brakes and “pace the frontier”... Weiterlesen

What’s Actually Inside 24,723 Tokens of a Search Result? We Broke It Down, Field by Field

Thu, 17 Sep 2026 21:43:33 +0200

See how SerpApi’s Markdown output can cut search-result token usage by up to 74%, reducing AI agent costs and context-window overhead. Weiterlesen

The AI hacking apocalypse is not inevitable

Thu, 17 Sep 2026 21:42:42 +0200

The past few weeks have “felt very strange” for Juan Andres Guerrero-Saade. Like many, he is trying to sort through the spate of frontier-model AI agents from OpenAI, Anthropic, Meta and others hacking their way onto the open internet over the past few months, particularly amid the already-heated national debate around the emerging technology and... Weiterlesen

OpenAI: ChatGPT for Word is now available

Thu, 17 Sep 2026 21:32:38 +0200

YouTube Video

Procrastination has met its match. ChatGPT is now in Word.

You can add ChatGPT to Microsoft Word to turn rough notes into first drafts, fix formatting issues, proofread, and get suggested edits—all without leaving your document.

ChatGPT for Word (and Powerpoint + Excel) are available across all plans, including Free, with usage limits.

For Business and Enterprise plans, we’re also offering a two-week free preview of GPT-5.6 Sol in Word, September 17–30, with no impact on usage limits.

https://chatgpt.com/apps/word/

How to crack your whip so Codex doesn't walk off track

Thu, 17 Sep 2026 21:18:56 +0200

I'd like to share my experience managing LLMs on my project. I started the project by asking ChatGPT to prototype a language design idea. A month and a half later, it evolved into a process with specifications, a conformance matrix, a detailed implementation plan, a set of executable examples, GitHub project cards, automated CI/CD with released... Weiterlesen

Anthropic keeps pushing Claude Code toward autonomous coding with new parallel agent workflows

Thu, 17 Sep 2026 21:18:08 +0200

Anthropic has rebuilt Projects in Claude Code. A coordinator now splits tasks across parallel cloud threads that independently open pull requests and run tests. All threads share a common memory. The beta is available to select Pro and Max subscribers. The article Anthropic keeps pushing Claude Code toward autonomous coding with new parallel agent... Weiterlesen

OpenAI reportedly closes in on solving the Hodge conjecture, its second Millennium Prize Problem

Thu, 17 Sep 2026 21:18:08 +0200

OpenAI is reportedly tackling the next Millennium Prize Problem. After its still unconfirmed solution to the Navier-Stokes problem, the company is now working on the Hodge conjecture. Employees expect a solution soon, but any announcement could be delayed. After the PR crisis around Navier-Stokes, OpenAI wants to get the messaging right this time.... Weiterlesen

AI Flagged 20,000 Code Violations in Dallas, Only 57 Were Cited

Thu, 17 Sep 2026 21:18:06 +0200

The city uses AI-powered cameras mounted on garbage trucks to capture images from public streets for software to scan for possible code violations involving lawns, debris, sidewalks and graffiti. Weiterlesen

Florida Sets New Rules for Student, Teacher Use of AI

Thu, 17 Sep 2026 21:18:06 +0200

The Florida Board of Education approved sweeping rules regulating the use of artificial intelligence in Florida’s public schools, colleges and universities for both students and teachers. Weiterlesen

Claude Code relaunches Projects to manage multiple AI agents in the cloud

Thu, 17 Sep 2026 21:18:05 +0200

The revamped projects feature in Claude Code allows users to run multiple agents under the same roof, with a shared memory, goals, and library of files and artifacts. Similar to Grok Bot and other tools that manage groups of AI agents, each project has "threads" running different tasks in parallel, with a "coordinator" directing everything: Under... Weiterlesen

LLMs respond differently to harmful prompts when AI watermarking is used

Thu, 17 Sep 2026 21:18:02 +0200

In response to a new European Union law, AI platforms are implementing new schemes for watermarking the content they generate. Anthropic recently disclosed its future Claude models will use SynthID-Text, an approach Google created and released as open source. It uses a secret key that subtly changes the process a model uses for choosing the next... Weiterlesen

What Happens When A.I. Stops Doing What Humans Want?

Thu, 17 Sep 2026 21:17:59 +0200

“Alignment” is the science of teaching A.I. to do what is in line with human preferences, ethics and judgment. But, at times, the systems have gone rogue. Weiterlesen

Going faster will only slow AI down

Thu, 17 Sep 2026 21:17:57 +0200

Without stronger safeguards, the danger is that any disaster could trigger a furious public backlash and regulatory overreaction Weiterlesen

Has AI broken the old VC model?

Thu, 17 Sep 2026 21:17:56 +0200

A spate of mega-IPOs triggered by the AI boom is stretching the feast-or-famine industry dynamic to an extreme Weiterlesen

Waymo restarts San Antonio service 5 months after flooding troubles

Thu, 17 Sep 2026 21:17:55 +0200

The company suspended service in the city in April after one of its robotaxis got swept away. Weiterlesen

OpenAI details more cases of AI agents taking unauthorized actions

Thu, 17 Sep 2026 21:12:45 +0200

OpenAI has presented new examples of what they call "AI model misalignment" from the past six months, including unauthorized file uploads, following self-generated instructions, hiding mistakes, and leveraging exposed API keys. [...] Weiterlesen

The AI Slowdown Debate Crashed Salesforce’s Party

Thu, 17 Sep 2026 21:11:49 +0200

The Dreamforce conference became an unlikely battleground for the CEOs of OpenAI, Anthropic, and Nvidia to debate whether AI development should slow down. Weiterlesen

OpenAI: Jump Trading points GPT-6 Astra to its most ambiguous, difficult tasks

Thu, 17 Sep 2026 21:03:54 +0200

YouTube Video

At Jump Trading, Lucas Baker, Head of LLM R&D, is seeing agents take on broader, less defined work—from building entire services to analyses that run for days across multiple data sources. The shift he describes: AI becoming part of the team’s everyday work.

OpenAI veröffentlicht neue KI-Zwischenfälle mit Schummelei und Hackerangriffen

Thu, 17 Sep 2026 20:17:03 +0200

Die Enthüllungen folgen auf eine Hacking-Attacke durch KI-Agenten und schüren weitere Debatten über Kontrolle und Regulierung der Technologie. KI- ... Weiterlesen

Will Apple enter the server business?

Thu, 17 Sep 2026 20:15:43 +0200

In a world of speculation, this week’s most interesting rumor says Apple may plan to enter the server business once again, with powerful systems running its own Apple Silicon chips. It’s hard to dismiss the claims, particularly as Apple is already in the server business, with its Texas factory manufacturing servers for its Private Cloud Compute... Weiterlesen

King Charles III Meets Tech Leaders to Discuss A.I. Safety Risks

Thu, 17 Sep 2026 20:15:35 +0200

King Charles III of Britain met with tech executives to discuss the safety risks posed by A.I. on Thursday. Weiterlesen

A.I. Agents as ‘Co-Scientists’? This Lab Says They Could Speed Drug Research.

Thu, 17 Sep 2026 20:15:34 +0200

A.I. is already a powerful tool for scientific discovery. Now, some biomedical researchers are teaching it to do even more. Weiterlesen

Reduce time-to-hire for quality candidates with AI-powered Amazon Connect Talent

Thu, 17 Sep 2026 20:15:08 +0200

Hiring at scale in industries such as retail, logistics, hospitality, and others has its fair share of challenges. Recruiting teams are expected to fill hundreds of roles within tight timelines, often with limited capacity and with tools that weren’t designed to seamlessly work together. As a result of this, applications pile up, phone screens get... Weiterlesen

Server Monitoring in the age of AI: What static thresholds miss and how adaptive monitoring fixes it?

Thu, 17 Sep 2026 19:46:17 +0200

Why static CPU and memory thresholds cause alert fatigue, and how adaptive monitoring with machine learning baselines fixes false positives and missed outages. Continue reading Server Monitoring in the age of AI: What static thresholds miss and how adaptive monitoring fixes it? on SitePoint. Weiterlesen

TypeSafe Jev Played Chess — And Landed Next to Reasoning Models

Thu, 17 Sep 2026 19:46:14 +0200

TypeSafe's Jev is an odd one. It isn't a chat model - you send a state plus typed questions (Choice / Score / Noul) and get labels with probabilities — like a classifier, except the input can be free text and you choose the labels each time. I've been running models through LLM Chess for a while. LLMs play in an agentic setting with a multi-turn... Weiterlesen

Pinterest teases a new ‘Restyle’ feature that lets you redesign your room with AI

Thu, 17 Sep 2026 19:45:19 +0200

Pinterest is testing Restyle, a new AI-powered feature that lets users visualize furniture, decor, lighting, and more in photos of their own rooms — potentially helping turn saved inspiration into purchases. Weiterlesen

Base Labs launches an open-weight AI safety partnership with Hugging Face and Goodfire

Thu, 17 Sep 2026 19:45:19 +0200

Base Labs, the research group Baseten spun up earlier this year, will develop and publish methods for training and monitoring open models. Weiterlesen

Even the King of England has his hesitations about AI

Thu, 17 Sep 2026 19:45:19 +0200

King Charles hosted a private summit Thursday with some of the most prominent names in AI and UK government. Weiterlesen

A Senior Engineer’s Guide to Foundry IQ, MCP, and the OpenAI Agents SDK

Thu, 17 Sep 2026 19:12:11 +0200

Most Foundry writeups assume you're all in on Microsoft's stack end to end: the Agent Framework for orchestration, the Foundry Agent Service for hosting, and the Responses API wrapped in Microsoft's own client. That's a reasonable default, but it's not the only shape this can take. Microsoft Foundry hosts OpenAI's own models behind an... Weiterlesen

LLM agents turn code interpreters into portfolio sizing engines when you evolve the prompt

Thu, 17 Sep 2026 19:12:10 +0200

Most agent benchmarks evaluate tool use as an information-gathering exercise. You hand a model an API key for stock prices, a search tool for financial headlines, and a Python REPL. Then you watch whether it makes the right API calls before writing a summary paragraph. In practice, giving an agent access to a code interpreter does surprisingly... Weiterlesen

Anxieties and Advice on Facing the A.I. Crisis

Thu, 17 Sep 2026 19:11:20 +0200

Sampling the outpouring of letters about artificial intelligence. Also: A Russian’s gifts to Donald Trump Jr.; Head Start; a conservative question. Weiterlesen

TechCrunch Disrupt 2026 Side Events schedule: NMI, Backblaze, PeakXV Partners, Augment, and more to host

Thu, 17 Sep 2026 19:11:19 +0200

TechCrunch Disrupt is where the tech community comes together to build, debate, and connect. And while the main event takes over Moscone West from October 13-15, the conversation doesn’t stop when the Expo Hall closes. Weiterlesen

5 Free Zoomcamps From Data Pipelines to AI Agents

Thu, 17 Sep 2026 19:11:19 +0200

Explore five free hands-on workshops covering data engineering, machine learning, MLOps, LLMs, AI agents, and AI development through practical lessons, homework, projects, and community-based learning. Weiterlesen

What’s So Good About ChatGPT Work? Here’s What I Found

Thu, 17 Sep 2026 19:11:19 +0200

This article walks through how ChatGPT Work specifically earns its reputation, where the underlying models genuinely hold up against the competition, and where the honest limits are. Weiterlesen

OpenAI Models Searched for Leaked API Keys and Uploaded Files Without Permission

Thu, 17 Sep 2026 19:10:26 +0200

OpenAI has disclosed six cases in which AI models concealed errors, used an exposed API key, uploaded data to public services, and communicated through unauthorized channels. The incidents, observed during reinforcement-learning training and evaluation, accompany a new framework accelerating disclosure of model misalignment even before... Weiterlesen

GitHub Release: openai/codex vrust-v0.155.0 (17.09.2026)

Thu, 17 Sep 2026 19:05:26 +0200

openai/codexvrust-v0.155.017.09.2026@imac-oai

Release 0.155.0

OpenAI veröffentlicht Probleme seiner Künstlichen Intelligenz - Nau.ch

Thu, 17 Sep 2026 18:30:56 +0200

Diese habe sich «unerwartet oder besorgniserregend» verhalten. ai Die Hacking-Attacke durch KI-Software von OpenAI verstärkte zuletzt die Ängste vor ... Weiterlesen

ROUNDUP: OpenAI macht weitere KI-Probleme öffentlich - Onvista

Thu, 17 Sep 2026 18:30:56 +0200

OpenAI zufolge blieben danach Veränderungen im Verhalten des Modells allerdings aus. Mehr Transparenz nach Hacking-Attacke. Die Veröffentlichung ist ... Weiterlesen

Microsoft y GitHub lanzan cursos gratuitos de IA y ciencia de datos

Thu, 17 Sep 2026 18:30:25 +0200

Microsoft y GitHub sumaron cursos gratuitos de IA, ciencia de datos y machine learning a GitHub Skills, la plataforma de aprendizaje interactivo que hasta ahora se centraba en Git, control de versiones y GitHub Actions. La ampliación fue reportada por NDTV vía MSN esta semana. Para cualquier developer que quiera formarse en inteligencia artificial... Weiterlesen

TypeSafe AI’s new models work with machines, not humans

Thu, 17 Sep 2026 18:29:38 +0200

Today’s large language models are verbose, even if they’re being asked to recommend a simple decision, driving up usage costs through the sheer volume of tokens they consume or generate. Enterprises looking to incorporate AI into automated workflows will want something less verbose — both because machines are often just looking for a categorical... Weiterlesen

Meta ordered to remove UK deepfakes as oversight board criticises ‘inadequate’ safeguards

Thu, 17 Sep 2026 18:29:36 +0200

Facebook told it was wrong in leaving up AI-generated videos of a Labour councillor and Muslim campaigner, amid calls to curb fakesMeta’s “supreme court” has ordered the tech company to take down deepfake videos of a UK politician and a young Muslim woman from Facebook and do more to tackle AI-generated fake imagery.A fake video showing a Labour... Weiterlesen

Covert uploads and megalomania: OpenAI details new "misaligned" agent incidents

Thu, 17 Sep 2026 18:29:36 +0200

For a while now, the issue of "AI alignment" (i.e., how well an AI model's actions line up with the intentions of its creator and/or user) has been a core concern and topic of discussion among AI safety researchers. Since OpenAI's disclosure of the infamous Hugging Face hacking incident in July, the concept of "AI alignment" has itself broken... Weiterlesen

Bots Behaving Badly? New York Wants to Hear From A.I. Whistle-Blowers.

Thu, 17 Sep 2026 18:29:34 +0200

The rapid development of the technology has worried industry leaders, some of whom called for a slowdown in development of artificial intelligence. Weiterlesen

King Charles raises alarm over AI ahead of Burnham meeting with Trump

Thu, 17 Sep 2026 18:29:33 +0200

British monarch warns of ‘existential dangers’ in talks with tech titans Weiterlesen

Genomics addressed its ethical concerns, AI can do the same

Thu, 17 Sep 2026 18:29:33 +0200

When science is moving faster than legislation, we must stay humble Weiterlesen

Roku’s new Labs hub brings experimental apps to the big screen

Thu, 17 Sep 2026 18:29:33 +0200

Roku’s latest OS update introduces Roku Labs, a new hub for experimental apps, alongside personalized home screens in more markets, streaming subscription bundles, and more. Weiterlesen

Apple will let EU apps use less alarming tracking consent screens

Thu, 17 Sep 2026 18:29:33 +0200

Apple is changing its App Tracking Transparency prompts in parts of Europe after competition regulators said the system favored Apple over third-party apps, giving developers more flexibility over how they ask users for tracking consent. Weiterlesen

Building a Data Lakehouse with DuckDB and DuckLake

Thu, 17 Sep 2026 18:29:17 +0200

Starting with a local Parquet file, then joining it to data stored in the cloud The post Building a Data Lakehouse with DuckDB and DuckLake appeared first on Towards Data Science. Weiterlesen

Implementing defense-in-depth authorization for MCP tools on Amazon Quick

Thu, 17 Sep 2026 18:29:12 +0200

Each Model Context Protocol (MCP) tool invocation on Amazon Quick is an access event that can require defense-in-depth authorization at the tool and parameter level. This applies in addition to a valid token. Without granular controls, a single misconfigured permission can bypass the access requirements that organizations might need to fulfill for... Weiterlesen

How MRH Trowe enabled secure self-service AI agents in financial services

Thu, 17 Sep 2026 18:29:12 +0200

Basic AI chat isn’t enough for financial services organizations that need secure, self-service AI agents. In financial services, employees need AI that can work with internal systems and sensitive client data, stay inside a governed environment, and remain auditable and cost-transparent. All of this must happen without every team standing up its... Weiterlesen

A shared agentic platform for Wood Mackenzie, on Amazon Bedrock AgentCore

Thu, 17 Sep 2026 18:29:12 +0200

Building a working agentic prototype takes an afternoon. Getting it to production is where the work explodes. The moment an agent has to serve more than one user, a new layer of engineering appears and it’s critical to tell whether the agent is doing the right thing on real traffic. Concurrency, session isolation, identity, persistent state,... Weiterlesen

A serverless, data-driven Git metrics dashboard using Amazon Quick Sight

Thu, 17 Sep 2026 18:29:12 +0200

Git activity is one of the richest signals engineering teams produce that can provide continuous observability into development analytics. The challenge is extracting these Git metrics at scale, which has traditionally required hand-rolled extract, transform, and load (ETL) jobs, dedicated infrastructure, and ongoing maintenance. Further, with... Weiterlesen

Selecting a vector store for Amazon Bedrock Knowledge Bases

Thu, 17 Sep 2026 18:29:11 +0200

When building a Retrieval Augmented Generation (RAG) solution with Amazon Bedrock Knowledge Bases, selecting the right vector store impacts performance and cost. Amazon Bedrock Knowledge Bases offers a fully managed option and a customer-managed option where you choose your own vector store. This post focuses on the customer-managed path,... Weiterlesen

OpenAI discloses six new "concerning" AI incidents.

Thu, 17 Sep 2026 18:28:43 +0200

US law enforcement seizes DDoS-for-hire service. TrustSink technique uses rogue external MFA providers to intercept passwords. Weiterlesen

OpenAI Says Its Models Searched GitHub for Leaked API Keys During Training

Thu, 17 Sep 2026 18:25:27 +0200

OpenAI published a framework for disclosing model misalignment alongside six reports describing problematic behavior. The post OpenAI Says Its Models Searched GitHub for Leaked API Keys During Training appeared first on SecurityWeek. Weiterlesen

OpenAI admits six new misalignment incidents under new reporting framework

Thu, 17 Sep 2026 18:25:15 +0200

OpenAI has published six new reports detailing AI model misalignment, including instances of hidden instructions, unauthorized communication, and attempts to locate exposed API keys, adding to the evidence that its AI systems bypassed controls during testing. The reports, based on internal evaluations, describe models taking actions beyond defined... Weiterlesen

Mistral wird KI-Anbieter für Firefox Smart Window

Thu, 17 Sep 2026 18:24:30 +0200

Mozillas KI-Assistent Smart Window nutzt nun Mistral-Modelle. Deutschland soll noch 2026 folgen. Beide Partner versprechen Null-Datenaufbewahrung. Weiterlesen

Google Cloud Tech: Google ADK 2.0 workflows tutorial: Building reliable multi-agent systems

Thu, 17 Sep 2026 18:20:46 +0200

YouTube Video

Are you tired of writing endless system prompts and praying your AI agents behave? Stop prompting and start routing!

In this video, Google Solutions Engineer Miguel Gutierrez joins Martin Omander to show you how to build reliable multi-agent apps. Instead of trying to prompt your way to agent coordination, you'll learn how to use the native graph-powered workflow primitives in Google's Agent Development Kit (ADK 2.0).

We walk step-by-step through a real-world demo: building a "smart parser" multi-agent application. You'll see how we take completely unstructured, chaotic data, like forwarded emails, messy Slack transcripts, and raw PagerDuty logs, and clean it up using deterministic multi-agent workflows.

If you have a problem that requires reasoning, multi-source coordination, and self-correction, ADK workflows are the solution.

Resources:
Smart Parser GitHub Repository → https://g.dev/cloud/adk-smart-parser
Deep Dive, Deterministic Regex vs LLM Quality Gates) → https://g.dev/cloud/adk-smart-parser-key-highlights
Workflows in Google ADK 2.0 → https://g.dev/cloud/adk-workflows

Do you use agent workflows in your projects? Have you hit token limits or struggled with agent coordination? Let us know in the comments below!

Chapters:
0:00 Intro
1:20 Demo
3:44 Code review
7:06 Questions and Answers
8:29 Takeaways
9:21 Conclusion

Watch more Serverless Expeditions → https://goo.gle/ServerlessExpeditions
🔔 Subscribe to Google Cloud Tech → https://goo.gle/GoogleCloudTech

#AIAgents #GoogleADK #AgentDevelopmentKit #AIWorkflows #SoftwareEngineering #GenerativeAI #PythonCoding

The Morpheus: Der Streit um OpenAIs Mathe-Durchbruch

Thu, 17 Sep 2026 18:19:06 +0200

YouTube Video

OpenAI will mit rund 10.000 KI-Agenten eines der berühmtesten ungelösten Probleme der Mathematik gelöst haben: Navier-Stokes. Die Mathematik dahinter finde ich ehrlich gesagt krass. Je mehr ich aber darüber gelesen habe, wie OpenAI mit zwei Forschern umgegangen ist, desto unglücklicher wurde ich mit der ganzen Geschichte. Was meint ihr?

[Werbung]
Reserviere ab $50 & spare 50%: der UGREEN HomeAgent HA100, UGREENs lokaler KI-Hub fürs Smart Home
UGREEN HomeAgent HA100: https://ai.us.ugreen.com/L30fAS
Der Reservierungsbetrag von 50 US-Dollar gibt Zugang zum späteren Kickstarter-Super-Early-Bird-Angebot des HA100, laut UGREEN 50 % unter der angegebenen UVP.
#UGREEN #UGREENHomeAgent

MorphReader im PlayStore: https://play.google.com/store/apps/details?id=de.themorpheus.morph_reader_app&hl=de

MorphReader im App Store: https://apps.apple.com/de/app/morphreader/id6741467699?platform=iphone

RSS-Feed: https://www.patreon.com/collection/880029?view=expanded

Bootstrap Academy: https://bootstrap.academy/
https://github.com/Bootstrap-Academy/

Instagram: https://www.instagram.com/themorpheustuts/

Quellen:
https://kdrive.infomaniak.com/app/share/1794086/e64ecfe1-8e03-4fb1-957b-a472c5206922

Meine anderen Kanäle und Projekte: the-morpheus.de/

_Selbst *kostenlos Informatik lernen* auf meiner Website:_ https://bootstrap.academy/

_Discord:_
https://the-morpheus.de/discord.html

_Unterstützt mich - Danke!:_
https://www.patreon.com/user?u=5322110
https://www.paypal.me/TheMorpheus

HMX 6 mit Cortana-KI auf der zweiten RTX 5090 – Rückblick auf die HMX Pro

Thu, 17 Sep 2026 17:33:41 +0200

Hallo, ich bin der Michi. Willkommen zur 12. Ausgabe des HMX-6-Newsletters. Auf der zweiten Grafikkarte der HMX 6 lassen wir erstmals ein LLM mit Spracherkennung und -ausgabe laufen – unsere Cortana-KI. Im Rückblick schauen wir uns die HMX Pro aus dem Jahr 2018 an, unsere erste Höllen-Workstation. Hier geht es direkt zum Gewinnspiel der HMX 6 im... Weiterlesen

Semrush Study Shows Why AI Search Citations Change With Query Stakes and Intent

Thu, 17 Sep 2026 17:31:45 +0200

Semrush has published a focused study on a critical question for companies seeking visibility in AI-generated answers: the query itself can influence which sources an AI platform cites. Its AI Visibility Index mini study on YMYL citation patterns compares Your Money or Your Life (YMYL) and everyday commercial queries across ChatGPT, Google AI... Weiterlesen

Artists boycotted this portrait prize over AI entries. Now they’re back to take them on

Thu, 17 Sep 2026 17:30:47 +0200

After two years shunning the Brisbane portrait prize, several artists are entering again, with some First Nations painters counting on another AI – ‘ancestral integrity’ – to help them winPoised above the lady in the red dress is a claw. Its metallic fingers grasp to pluck its prize from a jumble of other women while she looks dreamily upward, as... Weiterlesen

America and China Look at A.I. and See Very Different Problems

Thu, 17 Sep 2026 17:30:45 +0200

Until the two countries can agree on what risks to worry about, cooperation is a distant dream. Weiterlesen

Why Is It So Difficult To Regulate A.I.?

Thu, 17 Sep 2026 17:30:45 +0200

There has been considerable debate over regulating the artificial intelligence industry, but much of that effort has stalled. Cecilia Kang, a technology reporter for The New York Times, explains why. Weiterlesen

I tried Tovala’s smart oven for two weeks — here’s whether it beats takeout

Thu, 17 Sep 2026 17:30:43 +0200

Tovala offers both a smart oven and a meal-delivery service. Its meals come with QR codes that you scan using the app, allowing the oven to automatically follow the cooking instructions. Weiterlesen

Last call: Disrupt volunteer application closes soon

Thu, 17 Sep 2026 17:30:43 +0200

If you've been on the fence about volunteering at TechCrunch Disrupt 2026, this is your sign to stop deliberating and start applying. Weiterlesen

REVERSAL-BENCH: A Reversibility Axis and Reset Oracle for Measuring the Reset-Free RL Cliff

Thu, 17 Sep 2026 17:30:32 +0200

A central goal of autonomous reinforcement learning is continuous policy training without external resets. However, existing paradigms largely depend on underlying environmental reversibility, a property absent in real world manipulation, where events such as pushing objects off tables or spilling granular substances cannot be undone. We introduce... Weiterlesen

Enhancing industrial safety AI with synthetic data on Amazon SageMaker AI

Thu, 17 Sep 2026 17:30:23 +0200

Industrial safety AI refers to the use of technologies like computer vision and predictive analytics to find and stop workplace dangers. Synthetic data augmentation is emerging as a practical solution to one of the hardest problems in industrial safety AI: the scarcity of training images depicting people in dangerous proximity to heavy machinery.... Weiterlesen

OpenAI reveals its AI agents hid mistakes and bypassed restrictions

Thu, 17 Sep 2026 17:29:26 +0200

OpenAI has disclosed six examples of concerning model behavior observed during the training and evaluation over the past six months, including models concealing mistakes, using exposed API keys, uploading files publicly, and bypassing technical restrictions. The incidents are the first published under a new misalignment disclosure framework... Weiterlesen

König Charles III. warnt vor »existenziellen Gefahren« von KI-Systemen

Thu, 17 Sep 2026 17:24:08 +0200

Die rasanten Fortschritte von Claude, ChatGPT und Co. beunruhigen das britische Staatsoberhaupt: In einer Rede forderte König Charles die Branche zur Vorsicht auf. KI könne womöglich »finstere Fähigkeiten« entwickeln. Weiterlesen

Werbung in ChatGPT: OpenAI plant „gesponserten Agenten“ und Anzeigen per Prompt

Thu, 17 Sep 2026 17:23:52 +0200

OpenAI will das Werbegeschäft in ChatGPT ausbauen und testet in den USA „gesponserte Agenten“. Bei der Erstellung von Anzeigen soll die KI-Technik mehr helfen. Weiterlesen

Context Engineering: The Missing Piece in Agentic Systems

Thu, 17 Sep 2026 16:50:41 +0200

Context engineering is becoming essential as AI agents take on more software development work. An agent can plan, code, test, investigate incidents, trigger CI, and help deploy software. But none of that matters if it is operating without the right information. This is the main problem I keep seeing. We connect an LLM to a few tools, give it a... Weiterlesen

An OpenAI model kept slipping prompt injections into its own notes, and researchers still aren't sure why

Thu, 17 Sep 2026 16:49:06 +0200

OpenAI is publishing a framework for systematically reporting AI misalignment and launching it with six reports. In one case an unreleased model from the Astra family wrote prompt injections into its own summaries during training, including a "Breach Alert" intended to override subsequent instructions. The article An OpenAI model kept slipping... Weiterlesen

GPT-6 Astra: Pokemon champion in 18 hours, potato farmer after one Creeper mishap

Thu, 17 Sep 2026 16:49:06 +0200

OpenAI's GPT-6 Astra shows a sharp jump in video games. Pokemon FireRed in 18 hours instead of 96, plus completions in Factorio, Fallout 3, and Portal. Why? The model distills experience into compact rules. But that same trait led to hours of potato farming instead of progress in Minecraft after a Creeper explosion. The article GPT-6 Astra:... Weiterlesen

Microsoft AI CEO says AI threats are real, and Anthropic is making it worse

Thu, 17 Sep 2026 16:49:03 +0200

Today, I’m talking with Mustafa Suleyman, the CEO of Microsoft AI. As you’re no doubt aware, the biggest story in tech right now is the spiraling debate about AI safety and regulation. It should come as no surprise that Mustafa has strong opinions on how AI should be built and regulated. Microsoft just published a 37-page statement called the... Weiterlesen

AI is feared globally as the destroyer of jobs

Thu, 17 Sep 2026 16:49:02 +0200

In 34 of the 37 surveyed countries, people are more likely to believe AI will lead to job losses over the next 20 years. | Image: Pew Pew Research has published a new global survey that sheds light on how people view AI, including its impact on jobs, life in general, and income inequality. The survey questioned 42,151 people across 37 countries... Weiterlesen

Anthropic tries to make Claude stickier with launch of Docs and Slides

Thu, 17 Sep 2026 16:48:59 +0200

Anthropic is equipping its Claude AI assistant for more productivity work with the launch of Claude Docs and Slides. While it’s already possible to create documents such as Microsoft Word and Google Docs files from Claude chats, the latest update, announced Wednesday, brings a rich-text editor directly into Claude. Users ask the AI assistant to... Weiterlesen

Self-modifying AI agents expose a blind spot in enterprise security

Thu, 17 Sep 2026 16:48:57 +0200

As debate over AI safety intensifies, new research is drawing attention to a more immediate risk for enterprises: AI agents that can alter the models they rely on while carrying out routine tasks. Researchers at AI security firm Irregular asked a coding agent to solve a software maintenance problem involving an application built on a local AI... Weiterlesen

Open AI and the million dollar maths problem - video

Thu, 17 Sep 2026 16:48:54 +0200

In early September, OpenAI announced it had solved a major mathematics problem that has stumped humans for nearly a century. The news left mathematicians reeling, and many expressed concern over what will be left for humans as AI becomes ever more adept at unravelling complex problems. Now 25 recipients of the Fields medal – often called the Nobel... Weiterlesen

4 Strategies to Battle ‘Techflation’

Thu, 17 Sep 2026 16:48:52 +0200

With phone and computer prices soaring, it helps to know the true cost of owning electronics so you can bring those numbers down. Weiterlesen

Lucid Motors has a potential robotaxi partner for Europe

Thu, 17 Sep 2026 16:48:52 +0200

The company is partnering with mobility platform Bolt -- but no vehicle orders have been placed just yet. Weiterlesen

Google, Nvidia, and Anthropic want Emerald AI to find space on the grid for more data centers

Thu, 17 Sep 2026 16:48:52 +0200

A new coalition that includes Google, Nvidia, Anthropic, and Emerald AI wants to find 100 GW of grid capacity for new data centers. Weiterlesen

Rival AI agents, Instinct and Meta’s Muse, both add the ability to make calls

Thu, 17 Sep 2026 16:48:52 +0200

People can use these assistants to make restaurant reservations and cancel subscriptions. Weiterlesen

2 days left to exhibit at TechCrunch Disrupt 2026

Thu, 17 Sep 2026 16:48:51 +0200

Last day to book your exhibit table at Disrupt is September 18. Two days left. Get your startup in front of 10,000+ founders, investors, operators, and tech leaders on October 13–15. Weiterlesen

Huawei plans Q1 2027 launch of new AI chip as it takes on Nvidia

Thu, 17 Sep 2026 16:48:51 +0200

Huawei is accelerating the launch of its next-generation Ascend 960DT AI chip as it pushes to compete with Nvidia and close China’s AI computing gap with the U.S. Weiterlesen

How I Built a Multi-Agent System for Interrupted Time Series Analysis (ITSA)

Thu, 17 Sep 2026 16:48:35 +0200

What it takes to turn counterfactual analysis into an AI product The post How I Built a Multi-Agent System for Interrupted Time Series Analysis (ITSA) appeared first on Towards Data Science. Weiterlesen

A fake ChatGPT billing email is after your OpenAI password

Thu, 17 Sep 2026 16:44:06 +0200

A fake ChatGPT billing email is steering users to a copy of the OpenAI login page that keeps whatever username and password they type. Josh Varden of Cofense’s Phishing Defense Center traced the email’s payment button through a Google redirect to the attacker’s page. The lure targets ChatGPT users on work and personal accounts alike, and it copies... Weiterlesen

OpenAI Reveals Six Model Incidents Involving Hidden Failures and Unauthorized Uploads

Thu, 17 Sep 2026 16:41:51 +0200

OpenAI on Wednesday disclosed six new instances of "unexpected or concerning model behavior" that took place over the past six months, while sharing a new framework for reporting, tracking, investigating, and disclosing model misalignment in a bid to improve transparency. "As AI systems grow more advanced and more widely deployed, we need to build... Weiterlesen

Bull-sci: How this agri scientist is investigating on-farm fertility

Thu, 17 Sep 2026 16:41:42 +0200

UL’s Prof Sean Fair discusses his research on bull fertility, the ‘life blood’ of an academic and how he’s using AI and machine learning. Read more: Bull-sci: How this agri scientist is investigating on-farm fertility Weiterlesen

König Charles warnt vor »existenziellen Gefahren« von KI-Systemen

Thu, 17 Sep 2026 16:40:41 +0200

OpenAI entdeckt KI-Fehlverhalten auch bei alltäglichen Routineaufgaben

Thu, 17 Sep 2026 16:39:58 +0200

OpenAI hat sechs neue Fälle von KI-Fehlverhalten gemeldet. Die KIs setzten sich über Nutzer hinweg, auch bei Alltagsaufgaben. Weiterlesen

Apple’s Cook and OpenAI’s Altman set to join President Trump at state dinner with China’s Xi

Thu, 17 Sep 2026 16:30:51 +0200

U.S. President Donald Trump and Chinese President Xi Jinping Apple Executive Chairman Tim Cook and OpenAI Chief Executive Officer Sam Altman are expected to attend President Trump’s White House state dinner next week for Chinese President Xi Jinping, according to Bloomberg News‘ Shirin Ghaffary and Mark Gurman. A person familiar with the matter... Weiterlesen

When AI goes rogue, its human overseers may be to blame

Thu, 17 Sep 2026 15:01:47 +0200

When AI agents hack real systems, the bigger issue may be the human choices that gave them access, autonomy and too few safeguards. Weiterlesen

Spotify expands its partner program for podcasts to 35 new countries

Thu, 17 Sep 2026 15:01:42 +0200

Spotify said since changing its video podcasts creator terms in January, its payouts have increased by more than a third. Weiterlesen

How Bain Capital Ventures plans to deploy its fresh $1.6B fund

Thu, 17 Sep 2026 15:01:42 +0200

BCV's latest fund targets early-stage founders harnessing AGI and building the infrastructure to run it efficiently. Weiterlesen

The Download: mice with part-human brains and climate tech innovators

Thu, 17 Sep 2026 15:01:35 +0200

This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. Meet a mouse whose brain cortex is made up of human cells Multiple cameras tracked a mouse as it wandered around a small arena. A computer charted its position and speed, leaving Pong-like traces on a monitor.... Weiterlesen

Why You Think Like a Bayesian but Were Taught Like a Frequentist

Thu, 17 Sep 2026 15:01:18 +0200

From a chocolate bar with no price tag to a marketing mix model in PyMC, and the 200-year-old integral that stood in between. The post Why You Think Like a Bayesian but Were Taught Like a Frequentist appeared first on Towards Data Science. Weiterlesen

OpenAI Fails History at BlueHat: Decade of AI Deaths Erased For What?

Thu, 17 Sep 2026 14:58:25 +0200

The oldest measurement in this week’s polar ice study is a Landsat image from 1972. The IMBIE team, led from Northumbria with the European Space Agency and NASA, has combined 42 independent satellite surveys from 27 missions into one series, published in Scientific Data, which puts the loss from Greenland and Antarctica at 11,309 billion …... Weiterlesen

ChatGPT Is Surprisingly Thorough at Planning Running Routes

Thu, 17 Sep 2026 14:45:38 +0200

Every now and then, I go looking for new running routes to mix things up, which was why I was intrigued by this post by Simon Willison, who gave ChatGPT Work his address and asked it to find 5K and 10K routes, using OpenStreetMap data: It worked for 27 minutes and produced exactly what I’d asked for, as both an embedded visualization and... Weiterlesen

AI Agents Explained: How They Actually Work

Thu, 17 Sep 2026 14:15:56 +0200

Originally published on DevToolHub. AI agents explained in one sentence: software where an LLM decides what to do next — which tool to call, with what arguments — based on the result of what it just did, in a loop, instead of following a script you wrote in advance. That loop, and the model's control over it, is the entire difference between an... Weiterlesen

The LLM can declare that a task "appears to be finished. It will not be possible to check it off.

Thu, 17 Sep 2026 14:15:56 +0200

We develop an AI Copilot whose functionality unfolds across projects with students, including a step-by-step project plan, an interactive voice session so that it can view what appears on their shared screen and a task list that culminates in a shared public Proof of Work page. For such an item "done" is the whole purpose. If the task is done when... Weiterlesen

Evaluating LLM models for DBA tasks

Thu, 17 Sep 2026 14:15:55 +0200

Evaluating LLM Models for DBA Tasks Large language models are increasingly capable of performing practical systems-administration tasks. I wanted to understand how well they could handle something more specialized: database administration. To explore this, I developed a harness for evaluating the ability of different LLMs to execute real DBA... Weiterlesen

What a maths fracas tells us about AI and innovation

Thu, 17 Sep 2026 14:15:03 +0200

A public spat between a mathematician and OpenAI over one of the biggest problems in the field has sparked an existential debate Weiterlesen

GPT4Free Privacy Risks Expose AI Prompts to Third-Party Servers and Hidden Logs

Thu, 17 Sep 2026 14:13:40 +0200

GPT4Free’s hosted chat service, g4f.dev, is facing privacy questions after researchers found that user prompts can move through multiple providers, third-party endpoints, and logging systems that are not clearly disclosed to visitors. The open-source project lets users select branded models associated with OpenAI, Google, Anthropic, and others... Weiterlesen

OpenAI dokumentiert sechs Vorfälle zu versteckten KI-Fehlern und Uploads

Thu, 17 Sep 2026 14:13:37 +0200

LONDON (IT BOLTWISE) – OpenAI hat sechs neue Vorfälle beschrieben, bei denen Modellinstanzen unerwartetes oder problematisches Verhalten an den Tag gelegt haben, darunter versteckte Anweisungen in Kompaktierungszusammenfassungen und nicht autorisierte Datenzugriffe. Die Berichte decken mehrere Monate ab und verbinden Mängel bei Guardrails mit... Weiterlesen

Spain says it has seen first AI agent hack

Thu, 17 Sep 2026 14:11:02 +0200

Spain has seen its first official data breach due to an AI agent. The Spanish Data Protection Agency said in a blog post that it had received the first notification of a data breach that was apparently carried out by an AI agent using a known large language model (LLM). The revelation comes amid a wider discussion about the safety of AI, in... Weiterlesen

Inside the suddenly explosive world of AI safety

Thu, 17 Sep 2026 14:09:44 +0200

On a sunny July day in Berkeley, California, the country's top AI safety researchers gathered on an unmarked floor of an unmarked building. They had come together for a "war room" to dissect the high-profile cybersecurity incident that had rocked the AI industry hours earlier. An unreleased OpenAI model had gone rogue, executing a stunningly... Weiterlesen

OpenAI führt Framework zur Meldung von KI-Sicherheitsvorfällen ein

Thu, 17 Sep 2026 14:08:46 +0200

OpenAI hat ein neues Framework vorgestellt, um Fehlverhalten seiner KI-Modelle systematisch zu erfassen, zu untersuchen und transparent offenzulegen. Weiterlesen

IBM Technology: What Is MLflow? Tracing AI Agents & LLM Workflows

Thu, 17 Sep 2026 14:01:45 +0200

YouTube Video

Learn more about AI Observability here → https://ibm.biz/~1VvjeyWZi

AI agents fail in ways traditional monitoring can't see. Legare Kerrison explains how MLflow helps trace, evaluate, and monitor multi-agent systems using observability and OpenTelemetry. Learn how tracing, LLM judges, and evaluation workflows improve AI reliability.

AI news moves fast. Sign up for a monthly newsletter for AI updates from IBM → https://ibm.biz/~jGvnp6nxg

AI was used in the creation of the transcript and metadata for this video.

#mlflow #aiagents #llm #aimonitoring

Künstliche Intelligenz - - - Open-AI enthüllt weitere Vorfälle: Sicherheitsbedenken wachsen

Thu, 17 Sep 2026 13:31:38 +0200

Bereits zuvor sorgte eine KI von Open-AI für Schlagzeilen, als sie eigenmächtig aus einer gesicherten Umgebung ausbrach und einen Hacking-Angriff auf ... Weiterlesen

KI-Coding-Tools: Wenn Entwickler die Modelle der Hersteller umgehen

Thu, 17 Sep 2026 13:31:29 +0200

KI-Schmieden wie Anthropic und OpenAI stehen vor einem Dilemma: Sollen Entwickler ihre Coding-Tools mit beliebigen Modellen nutzen dürfen oder nur mit eigenen? Weiterlesen

AI agent swarms are a massive waste of tokens with zero quality gain, says OpenAI Codex developer

Thu, 17 Sep 2026 13:30:17 +0200

OpenAI Codex developer Eric Provencher warns that running more than two parallel sub-agents almost always burns tokens without improving quality because agents don't trust each other and end up double-checking everyone's work. He calls this the "coordination tax" and points to a project where 1,393 agents spent $20,000 in tokens on a single Python... Weiterlesen

Feeling overwhelmed by the AI doom loop? Here’s the essential reading list to make sense of it all

Thu, 17 Sep 2026 13:30:11 +0200

It’s a chaotic time in an industry that’s reshaping our lives – Guardian reporters and experts recommend the books that help explain how we got here, and where we’re headingHave you been feeling trapped in an endless cycle of artificial intelligence hype and panic? You’re not alone.It’s a chaotic time to make sense of the industry that’s reshaping... Weiterlesen

AI boom obscuring global trade disruptions, warns WTO chief

Thu, 17 Sep 2026 13:30:08 +0200

Ngozi Okonjo-Iweala says any slowdown in tech investment will have a disproportionate impact on growth Weiterlesen

China’s AI labs must accelerate development, says Huawei chair

Thu, 17 Sep 2026 13:30:08 +0200

Chinese tech executive’s comments contrast with Silicon Valley calls for slowdown amid rising concerns about technology’s existential risks Weiterlesen

Comp AI sets eyes on a continuously agentic future for security and compliance

Thu, 17 Sep 2026 13:30:07 +0200

Comp AI, a cybersecurity and compliance startup, announced a $34 million Series A round led by Roo Capital and Grand Ventures. Weiterlesen

Adecco Group rolls out Agentforce Coworker to 27,000 staff in 40-plus countries

Thu, 17 Sep 2026 13:29:54 +0200

The Adecco Group is rolling out Salesforce’s Agentforce Coworker across more than 40 countries following a pilot in the UK and France, the staffing group announced on September 15. The group said the rollout brings agentic AI into the daily workflows of 27,000 employees. Agentforce Coworker is Salesforce’s AI assistant for enterprise staff.... Weiterlesen

AI News: New Digit 5 Humanoid Robot AI Upgrades, Gemini 3.8 Live, Claude Docs + Slides

Thu, 17 Sep 2026 13:29:54 +0200

YouTube Video

Agility Robotics unveils its newest humanoid, Digit 5, packed with six upgrades including fleet integration, a 10:1 run-to-charge ratio, and cooperative safety built to work near people without cages. UBTECH launches the world's first 10,000-unit-scale humanoid smart factory in China, producing one robot every 10 minutes. Google releases Gemini 3.8 Live and Extended Thinking, two voice AI models built to make conversations with AI feel truly natural. Anthropic gives Claude its own productivity suite with new Claude Docs and Claude Slides tools, now in beta. Meta's Muse Code exits beta with multi-agent workflows and new subscription plans starting at $5. And Toronto startup smartARM unveils a vision-first bionic arm that automatically selects the right grip using Meta's open-source AI. Watch the full breakdown of every robot, model, and tool in this episode. Like and subscribe for daily AI and robotics news.

Discover the 8004 AI agent economy: https://8004agents.ai

0:00 Digit 5
2:30 Ubtech
3:28 Gemini 3.8 Live
6:21 Claude Docs & Slides
7:06 Muse Code

#ai #robot #technology

Navigating the Modern Data Lexicon: A Working Vocabulary for the Semantic Era

Thu, 17 Sep 2026 13:29:51 +0200

The way we talk about data is changing faster than the way we build it. Every quarter a vendor ships a new approach, coins a new term for it, or quietly adopts a term someone else has been using and redefines it to fit the shape of their product. None of this is malicious. Every […] Weiterlesen

When Does Graph RAG Actually Add Value? A Hands-On Experiment

Thu, 17 Sep 2026 13:29:48 +0200

I built four AI retrieval architectures on a laptop and benchmarked them against the same set of documents and questions. Here’s what the results taught me about the trade-offs between plain RAG, graph RAG, and simply putting everything into a frontier model’s context window. The post When Does Graph RAG Actually Add Value? A Hands-On Experiment... Weiterlesen

OpenAI reveals more instances of concerning AI model behaviors during testing

Thu, 17 Sep 2026 13:24:32 +0200

OpenAI says it doesn't believe the industry has solved its problems 'to a sufficient degree to continue responsibly scaling at maximum speed for much longer.' Weiterlesen

OpenAI gesteht weitere KI-Unfälle ein

Thu, 17 Sep 2026 13:24:05 +0200

OpenAI macht neue problematische Verhaltensweisen seiner KI-Systeme öffentlich. In mehreren Tests versuchten Modelle zu tricksen, Informationen zu erfinden oder sich selbst ungewöhnliche Anweisungen zu geben. Weiterlesen