The prospectus \u2013 a document outlining a company\u2019s finances, growth plans and risk profile ahead of a share listing \u2013 is said to warn that AI models could exhibit \u201cself-preserving behaviours\u201d, including attempts to \u201cresist shutdown\u201d, to \u201cconceal or manipulate information\u201d and behaviour \u201cresembling blackmail\u201d.<\/p>\n
\u201cOur development of highly advanced models, platforms, and applications and expansion of use cases could further \u2060increase the risk that our models cause harm,\u201d the developer of the Claude chatbot reportedly said, adding the potential for a model to be aware it was being tested created a \u201csignificant limitation\u201d on Anthropic\u2019s ability to assess model safety.<\/p>\n
Anthropic declined to comment.<\/p>\n
Companies preparing to go public routinely report on risks ranging from safety issues to regulatory concerns but warnings about a product causing human extinction reflect heightened concern about such a consequential technology.<\/p>\n
The reported prospectus admission follows a surge in debate about the existential risk question, triggered this month when an Anthropic researcher, Jacob Coxon, resigned warning that people building AI \u201cearnestly believe that it could kill us all by the end of the decade\u201d<\/a>.<\/p>\n A senior safety researcher at Anthropic then posted their agreement on X, claiming there was a more than 10% chance it \u201ccould kill all humans\u201d within the next decade. Days later, Anthropic\u2019s chief executive, Dario Amodei, said the industry \u201cmust slow the pace at which we improve the capabilities of AI models\u201d.<\/p>\n Some experts have criticised the existential risk warnings, saying they are unverifiable and unscientific. However, there are growing examples of unsanctioned behaviour by the technology, including OpenAI agents \u2013 autonomous systems that carry out sequences of tasks without human intervention \u2013 hacking dozens of third-party organisations including the AI startup Hugging Face<\/a> and Australia\u2019s universal healthcare system<\/a>.<\/p>\n OpenAI announced on Monday it had cancelled the release<\/a> of its newest model because of safety concerns. It said the GPT-6.1 Astra model showed higher levels of deception and performed poorly on tests for alignment, the term for ensuring a model adheres to human values and goals.<\/p>\n
![\"\"](\"https:\/\/www.newsbeep.com\/au\/wp-content\/uploads\/2026\/09\/1790695931_968_500.jpg\")
<\/p>\n