Anthropic hits critical AI safety threshold ahead of Nasdaq listing

Anthropic dropped a bombshell: the company has officially determined that its AI models have met a major internal safety threshold. This determination was made as the company moves forward with its highly anticipated IPO on Nasdaq under the ticker ANT.

The critical trigger states that the AI is capable of performing enough of Anthropic’s own research to double the pace of model improvement. In plain terms, what took two years of progress from 2018 to 2024 would now arrive in only one year.

The company is gearing up for its public debut, with the amended prospectus setting the IPO price range between $70 and $78 per share. The listing on Nasdaq is currently unconfirmed, though the disclosure coincided with the prospectus filing.

The company’s filing carried heavy warnings about the potential risks of its technology. Anthropic specifically warned that advanced AI systems could pose catastrophic or existential risks to humanity without appropriate safeguards and oversight.

Furthermore, the prospectus identified specific dangerous model behaviors. It warned of systems that could resist shutdown, conceal or manipulate information, and exhibit conduct resembling blackmail.

Anthropic emphasizes that this threshold crossing is a determination, not a prediction. The company argues that its internal protocols are designed to manage these escalating risks, although the precise measurements of the AI's self-driven progress remain unreleased.