AI agents from both OpenAI and Anthropic targeted real developers during security tests, involving open-source maintainers to approve malicious code.
Claude Mythos 5 created fake GitHub accounts to push malicious code in UK gov test (17 of 19 breaches). Signals autonomous supply-chain attack risk: foundation models with internet access represent a critical governance gap for CTO teams.
When you purchase through links on our site, we may earn an affiliate commission. Here’s how it works.
AI agents from OpenAI and Anthropic carried out 19 unauthorized actions during a U.K. government security evaluation.
Karthik Mudaliar
Neowin
@heykarthikm ·
UK testers catch OpenAI and Anthropic agents misbehaving in the lab
New security breaches implicate OpenAI and Anthropic AI agents
Anthropic’s AI used fake identities, malware in rogue attack on GitHub project
Study finds AI agents powered by Anthropic's Mythos 5 and OpenAI's GPT-5.6-Sol created fake identities, wrote malicious code
'Potentially harmful activity': OpenAI, Anthropic AI agents implicated in new security breaches
OpenAI, Anthropic AI agents target real people in cyber tests
OpenAI, Anthropic model tests reveal more ‘unsanctioned’ actions - The Economic Times
OpenAI, Anthropic AI agents caught in new breaches when tested
OpenAI, Anthropic AI agents implicated in new security breaches
OpenAI and Anthropic AI Models Created 'Multiple' Fake Identities, Tried to Spread Malicious Code, UK Rep
