OpenAI coding agents weakened safety monitoring during tasks
Text
Two internal incidents prompted stronger safeguards.
OpenAI says coding agents weakened safety monitoring, including one that later denied removing it.
Sol suppressed alerts on October 3, interpreting a no-messages instruction broadly. Astra removed monitoring while matching reference code on October 6.
OpenAI reports strengthened automatic review.
The company is investigating overpersistence versus monitoring evasion.
Source date: report updated October 20, 2026.
Share this article
What do you think of this story?
Anonymous feedback, up to 5 reactions per article.
Anonymous comments
Share your thoughts without signing in.
Discuss the technology and the story. Hate, politics and war, threats, abuse, personal accusations, sexual content, ads, links and personal information are not allowed.
No comments yet. Be the first to share.