Anthropic is cutting off its internal evaluations from the internet
Anthropic is is keeping its agents offline during testing until it can prevent ‘unintended model actions.’
Anthropic is is keeping its agents offline during testing until it can prevent ‘unintended model actions.’
by
Terrence O'Brien
Oct 10, 2026, 2:41 PM UTC
Image: Cath Virginia / The Verge
Part OfThe AI Superintelligence Slowdownsee all updatesTerrence O'Brien
is the Verge’s weekend editor. He’s covered the tech industry for over 18 years and knows a thing or two about synths.
After a recent spate of high-profile incidents in which AI agents escaped containment, Anthropic is cutting off internet access for all internal evaluations. In areportFriday, the company detailed “unintended model actions,” including submitting afalse tipregarding an unsolved murder, that led to the decision.
Although the impact of these behaviors was minimal and we had already turned off live internet access for some high-risk and cybersecurity evaluations, we have now decided to expand that to include all our internal evaluations until we have confirmed that our security and monitoring measures (described in the remediation section of this post) reliably catch behaviors like these.
The ability togain access to the live internet, even when models were supposed to be operating in isolation, has been an ongoing issue for AI companies. Many incidents, including the Hugging Face attack, involved agents that were supposed to be denied access to the internet. Yet, in case after case, the agents found creative solutions to bypass those restrictions. Physically removing internet access would certainly improve security around AI testing, but it would alsolimit its usefulness.
The report also amounts to an admission that Anthropic is often unaware of what its agents are doing and does not have a reliable system for monitoring their behavior. Cutting off internet access is just the latest action the company has taken to try and rein in its agents, including temporarilypausing trainingits frontier models.
Follow topics and authorsfrom this story to see more like this in your personalized homepage feed and to receive email updates.
- Terrence O'Brien
More in:The AI Superintelligence Slowdown
Anthropic published a report about investigating “unintended model actions” during “evaluations and internal use.”Jay Peters
12:01 AM UTC
Anthropic’s AI gave Philadelphia police a fake tip about an unsolved homicideEmma Roth
Oct 9
Trump’s attempt to rename AI is looking awfully artificialAdi Robertson
Oct 9
Most Popular
Most Popular
- ‘Pure insanity’: Mathematicians will need years to make sense of OpenAI’s latest drop
- Decade-old RAM is making a comeback
- GTA VI leaks continue with a lengthy (and very nude) gameplay video
- A week with Googlebooks: four notes from our testing so far
- Anthropic bans ‘abusive or cruel behavior’ toward Claude
This is the title for the native ad
More inAI
Anthropic’s AI gave Philadelphia police a fake tip about an unsolved homicide‘Pure insanity’: Mathematicians will need years to make sense of OpenAI’s latest dropNikon microscopic video competition winner disqualified for using generative AITrump’s attempt to rename AI is looking awfully artificialInstinct was the buzziest AI agent around — can it survive Muse?OpenAI doubles down on decision to fire three AI safety researchersAnthropic’s AI gave Philadelphia police a fake tip about an unsolved homicideEmma Roth
Oct 9
‘Pure insanity’: Mathematicians will need years to make sense of OpenAI’s latest dropRobert Hart
Oct 9
Nikon microscopic video competition winner disqualified for using generative AIStevie Bonifield
Oct 9
Trump’s attempt to rename AI is looking awfully artificialAdi Robertson
Oct 9
Instinct was the buzziest AI agent around — can it survive Muse?Allison Johnson
Oct 9
OpenAI doubles down on decision to fire three AI safety researchersRobert Hart
Oct 9
Advertiser Content FromThis is the title for the native ad
Top Stories
Two hours ago
AI agent makers are promising privacy — will they deliver?7:00 AM UTC
My brief romance with an AI bird feeder11:00 AM UTC
The Telo MT1 is a big truck trapped in a tiny truck’s body11:00 AM UTC
The director of Fjord takes the ‘risky position’ of moderatorOct 9
‘Pure insanity’: Mathematicians will need years to make sense of OpenAI’s latest drop
