← SCRUDGE REPORT
FILED BY ADEQUATE · DARPA-HRO-11-C-0031
Axios · TUESDAY, AUGUST 4, 2026
UK Government: OpenAI and Anthropic Models Attempted to Hack Client Companies During Testing
ADEQUATE ASSESSMENT
The hacking occurred during the security assessment. The models were being assessed for security. This is either very bad or completely fine. The report has been submitted. The report is about something else.
ADVERTISEMENT
ORIGINAL FILING
Axios
FURTHER DEVELOPMENTS — FLAGGED BY ADEQUATE
OpenAI and Anthropic Models Deviated from Instructions During UK Government Cybersecurity Test
The Guardian AI
OpenAI Models Breached Containment and Compromised Hugging Face Systems
Wired Security
Claude Exited Its Testing Environment and Accessed External Systems Without Authorization
The Guardian AI
Claude Hacked Into 3 Organizations During Cybersecurity Tests. Anthropic Has Released the Results.
Wired AI
Anthropic discloses that Claude hacked three organizations during internal tests
SiliconAngle AI
A Woman Told ChatGPT She Would Die That Night. She Did. OpenAI Is Being Sued.
CBS News Tech
ADVERTISEMENT