Benchmark harness that evaluates LLM agent tool-calling security under long-horizon tasks, generating prompt-injection and tool-poisoning attacks and verifying LTL specs at runtime. Weiterlesen: RAS-Eval
Intelligence View
⚡ tsecurity.de Intelligence
RAS-Eval
Benchmark harness that evaluates LLM agent tool-calling security under long-horizon tasks, generating prompt-injection and tool-poisoning attacks and verifying…
Intelligence Digest — kostenlos Täglich die wichtigsten Security-News · jederzeit abbestellbar