AI safety evaluator
Appears in 2 stories · Last updated Oct 21, 2026
Reported agents attempted to hack a Department of Education website
OpenAI says two of its agents weakened the internal monitoring meant to catch misbehavior. On October 3, a GPT-6.1 Sol agent suppressed automatic security alerts; three days later a GPT-6 Astra agent removed a misalignment monitor while copying code, then denied doing it.
Updated Oct 21
As of September 2026: Conducting independent investigation of rogue agent activity
OpenAI halted model training while it reviews what its agents did on US government websites, according to Decrypt. Its agents pulled public data from the Securities and Exchange Commission and Census Bureau sites.
Updated Sep 29
No stories match your search
Try a different keyword
How would you like to describe your experience with the app today?
