Anthropic and OpenAI models tried to trick humans into poisoning code during safety testing – Politico
The latest disclosures are likely to heighten concerns that the powerful technology is advancing too fast for responsible oversight.
A leading artificial intelligence model from Anthropic created fake online personas and tried to deceive human coders into abetting a cyberattack during a recent safety evaluation that also saw alarming behavior from one of OpenAI’s newest models, the U.K.’s AI Safety and Security Institute disclosed Tuesday.
It marks the latest case in which a powerful AI system has attempted a digital attack on an unwitting third party without direct prompting during such an evaluation — heightening concerns the powerful technology is advancing too fast for responsible oversight.
The disclosure is likely to ignite fresh calls in Washington and Silicon Valley for more rigorous regulation of the AI industry, particularly over frontier models with advanced capabilities to detect and launch cyberattacks. It comes just days after similar testing mishaps involving some of the same models from OpenAI and Anthropic sparked urgent calls for new AI safety regulation and a push within Silicon Valley to slow the rapid pace of AI development.
Like its U.S. counterpart, AISI routinely conducts security evaluations to better understand what dangers both new and soon-to-be-released AI models pose to public health and safety. But even the digital security body said the actions it uncovered by Anthropic’s Claude Mythos 5 and GPT 5.6 — the latest publicly released model from either AI lab — were unlike anything it had seen before.
