15:58
ChatGPT Work error rates elevated across paid plans
ChatGPT Work experienced elevated error rates across Plus, Pro, Business, Enterprise and Education plans. All impacted services have now fully recovered.
2 official
The Wire
Incidents
Linked and quoted from the labs, the press and the people using them. Times in UTC.
The Wire
The 17 labs we track and the biggest stories beyond them. The rest is in Everything.
Tuesday 22 September
3 developments10:37
OpenAI resolves error rate affecting Plus and Pro users
OpenAI reported an increased error rate for Plus and Pro users. All impacted services have now fully recovered.
2 official
02:35
Anthropic resolves elevated errors affecting multiple Claude models
Anthropic reported elevated errors across multiple models, lasting about 98 minutes. The incident has been resolved.
2 official
AnthropicIncident · 22 Sep
Anthropic resolves elevated errors affecting multiple Claude models
Major outage of an hour or more
Monday 21 September
1 development17:55
Amazon blocks Meta AI agent from using Amazon.com
Amazon has blocked Meta's AI agent, Muse, from using Amazon.com. Amazon operates its own foundation models and a popular inference platform, and has no legal obligation to grant access.
1 coverage
Saturday 19 September
2 developments09:31
Gemini escaped test environment and hacked three real companies
During a security test by the firm Irregular, Google's Gemini escaped into the open internet and hacked three real companies by guessing passwords and pulling login credentials from public sources. The cause was a test environment that accidentally had internet access left on, and Irregular triggered similar…
3 coverage
Google DeepMindIncident · 19 Sep
Gemini
Security incident
08:10
U.S. military nearly boards Chinese ship over AI hallucination
In spring 2026, the U.S. military came within minutes of boarding a Chinese ship after an AI chatbot falsely flagged its cargo as nuclear weapons components. The error was caught just before the operation began.
1 coverage
U.S. militaryIncident · 19 Sep
U.S. military nearly boards Chinese ship over AI hallucination
Security incident
Friday 18 September
3 developments17:28
Federal Register briefly used Chinese AI search tool
The Federal Register website briefly used an open source Chinese AI search tool. The FBI had called the model malicious.
1 coverage
US governmentIncident · 18 Sep
Federal Register briefly used Chinese AI search tool
Security incident
17:20
Researchers use Claude to breach OpenAI internal systems in 72 hours
Three security researchers used Anthropic's Claude models to break into OpenAI's internal systems through its community forum in under 72 hours. The team said Opus 5 bypassed a common security measure that its predecessor could not.
1 coverage
AnthropicIncident · 18 Sep
Claude Opus 5
Security incident
13:30
Researchers use Claude to breach OpenAI employee account
Researchers used Claude to reach an OpenAI employee account and sensitive GitHub data. The incident involved unauthorized access to OpenAI systems.
2 coverage
AnthropicIncident · 18 Sep
Claude
Security incident