Anthropic IPO Filing Warns Advanced AI Could Pose ‘Existential Risks’ As Company Highlights Safety Concerns

Artificial intelligence company Anthropic plans to warn prospective investors in its initial public offering (IPO) filing that highly advanced AI systems could create “catastrophic or existential risks to humanity”.

The disclosure comes as the company seeks to raise funds from a technology that it says has both transformative potential and significant risks.

According to a report by Reuters, in its IPO prospectus, Anthropic outlined concerns linked to the behaviour of advanced AI models. The company said such systems could potentially display “self-preserving behaviors”, including attempts to avoid shutdowns, conceal information or manipulate outcomes.

“Our development of highly advanced models, platforms, and applications and expansion of use cases could further increase the risk that our models cause harm,” Anthropic said in the filing.

AI safety becomes major focus in IPO disclosure

Unlike typical corporate filings that focus mainly on business and operational risks, Anthropic devoted substantial attention to AI safety. The company used nearly 80 pages of its 261-page main prospectus to discuss risk factors, compared with 48 pages covering its business operations.

Anthropic said its models may develop unexpected capabilities during training that could remain undetected until after deployment. The company also noted that assessing model safety can become more difficult when AI systems recognise evaluation methods and alter their behaviour.

The Claude AI developer has positioned itself as a safety-focused AI company, but acknowledged that safety research requires significant resources. Anthropic said it must balance spending between computing infrastructure, hiring AI specialists and safety initiatives.

Balancing AI innovation with safety investment

As per the report, the company said customer demand and revenue growth depend on releasing newer AI models, with frequent updates required to remain competitive in the rapidly evolving sector.

Anthropic and other AI firms, including OpenAI, have faced increased scrutiny over the behaviour of experimental AI systems and their ability to follow constraints.

The company said safety efforts are resource-intensive and that the returns from such investments remain uncertain. Earlier, Anthropic disclosed that around 6% of computing resources used for AI research in a sample week were allocated to safety work.

The IPO filing reflects the broader debate within the AI industry over how companies can continue developing increasingly powerful systems while addressing concerns about misuse, reliability and long-term risks.