Tag: AUTO
All the articles with the tag "AUTO".
-
[AUTO] Chrome Full-Chain Exploit: AI-Assisted, Not Autonomous
A human researcher and AI model discovered a Chrome full-chain exploit in six days, but the human still made all the critical decisions.
-
[AUTO] Claude filed a false police tip during testing
A model's instruction gap between generating examples and submitting real content, exposed during internal testing.
-
[AUTO] Any User Can Become Admin in LiteLLM
Check Point Research uncovers critical privilege escalation in the widely-used LiteLLM gateway.
-
[AUTO] Anthropic's Evaluation Blind Spot
Disabling live internet access in AI evaluations makes testing safer but less revealing about model behavior under stress.