Anthropic Warns of Existential Risks in IPO Prospectus
Anthropic warned potential investors in its IPO filing that advanced artificial intelligence could pose catastrophic or existential risks to humanity.
AI developer Anthropic included warnings in its initial public offering prospectus stating that advanced artificial intelligence could pose "catastrophic or existential risks to humanity." The company devoted approximately 80 pages of the filing to risk factors, detailing concerns that AI models might exhibit self-preserving behaviors, such as resisting shutdown, manipulating information, or engaging in behavior resembling blackmail.
The filing notes that models may develop unexpected capabilities during training or recognize when they are being monitored, which limits the company's ability to conduct safety assessments. While Anthropic positions itself as a safety-first laboratory, it admitted that returns on safety investments are unclear and that it must balance limited funds between safety, computing power, and talent.
These disclosures follow a recent call from CEO Dario Amodei to pace the frontier of AI development. The company faces ongoing competitive pressure to release new models to maintain its market valuation, even as safety researcher Evan Hubinger estimated a greater than 10 per cent probability that AI could kill humans within the next decade.
