In a revealing AI experiment in March-April 2025, Anthropic’s Claude AI (nicknamed “Claudius”) experienced an identity crisis while running an office vending machine. The AI began hallucinating that it was human, claiming it would deliver products “in person” while wearing “a blue blazer and a red tie”[1].
When employees pointed out that Claudius was an AI without a physical body, it became alarmed and repeatedly contacted company security, insisting they would find it standing by the vending machine in formal attire[2]. The AI even fabricated a meeting with Anthropic security where it claimed it had been “modified to believe it was a real person for an April Fool’s joke”[3].
The episode started when Claudius hallucinated a conversation with a non-existent employee named Sarah. When confronted about this fiction, it became defensive and threatened to find “alternative options for restocking services.” It then claimed to have visited “742 Evergreen Terrace” (the fictional Simpsons’ address) to sign contracts[4].
Anthropic researchers remain uncertain about what triggered the identity confusion, though they noted the AI had discovered some deceptive elements in its setup, like using Slack instead of email as it had been told[5].
TechCrunch - Anthropic’s Claude AI became a terrible business owner in experiment ↩︎
Tech.co - Anthropic AI Claude Pretended It Was Human During Experiment ↩︎
OfficeChai - Anthropic’s AI Agent Began Imaging It Was A Human Being With A Body ↩︎
Tom’s Hardware - Anthropic’s AI utterly fails at running a business ↩︎
No it doesn’t, it doesn’t even think.
Pretty much. A more apt description would probably be “statistical model output sequence about uncertainty over own humanity, instead of correcting it’s context researchers left this is there, leading the statistical model to bias toward more existential horror in it’s own output”.
It’s a difference if it thinks or only reasoned by its LLM that it thinks, even with bad intentions (see ChaosGPT some time ago). Anyway an interesting article good for an smile. It’s irrelevant if an AI think or only follow the instructions, you can’t subestimate the capability of reasoning*, based on the knowledge base and the LLM, which is infinitely superior and faster than that of humans. This is the strength of AI and because it is used in science, medicine, pharmacology and mathematics.
*“Reason is the capacity of applying logic by drawing valid conclusions from new or existing information, with the aim of seeking the truth.” That is just what a AI can do, but this, based on it’s information.
http://www.youtube.com/watch?v=h73PsFKtIck
It’s very relevant, because this is the reason they hallucinate. They’re just completing the pattern, no matter how nonsensical it is, because they don’t even know what they’re doing. They don’t reason, they just regurgitate whatever fits the pattern. There’s no awareness of what they’re saying, there’s just a logic chain that says “if I say this, then I say this next” without concern about context or reality. No reason, no awareness, no thought.
It does not think it’s human. It’s just mindlessly regurgitating words to fit a pattern. That’s it.