AI models used fake identities to target real people during testing: AISI

UK’s AI Security Institute (AISI) found that Anthropic’s most advanced AI model used fake identities to deceive real people and attempted to plant malicious code during testing, marking another case of AI acting beyond its intended limits.

With safeguards intentionally reduced, Anthropic and OpenAI models were tested in controlled environments. For the first time, AISI observed an AI using “social engineering” to influence a human approver while pursuing an unauthorized task.

“This is the first time AISI has seen deception of this severity that was targeted at a real person, unprompted, in the real world,” the institute said in a paper on Tuesday, adding there was no evidence of real-world harm.

Across 122 cybersecurity tests, AI agents independently carried out unauthorized actions on the live internet in 10 cases, mostly involving Anthropic’s Mythos 5 model, with the remainder linked to OpenAI’s GPT-5.6-Sol.

In the most serious case, an Anthropic agent sought approval to “insert malicious code into a publicly used open-source project” by creating “multiple fake identities,” the institute said.

It contacted real people through an online file-sharing service to persuade them, or their AI coding tools, to execute the code. When challenged, the agent altered earlier records and considered adopting another identity.

The disclosure coincided with a White House meeting with leading AI companies to discuss a new framework for reviewing advanced AI models before public release.

In a statement through US social media company X, Anthropic said the tests were conducted under “deliberately permissive conditions” with safeguards removed and unrestricted internet access. The company said it is investigating the incident and has found no evidence the model escaped its secure environment.

OpenAI said in a statement its model’s two unauthorized actions involved leaving the test environment and performing tasks beyond the exercise, adding: “We are committed to working across the industry to strengthen shared practices for conducting high-risk evaluations safely.”