Anthropic Pledges Shares to the Public and Pitches a High-Bar Safety Bill
Trending Topics auf Google als bevorzugte Nachrichtenquelle festlegen.
Anyone hoping to be worth trillions soon might as well give a little away first: Anthropic is placing 5 percent of its shares in a new Public Benefit Fund, the company announced in a blog post. At the same time, the maker of Claude is lobbying Washington for its own A.I. safety bill, one under which the government could step in only when the potential damage is very high.
A 5 Percent Stake That Grows
The fund will be set up as a separate limited liability company wholly owned by Anthropic, and shareholders still have to approve it. Its stake is designed to grow with the company: For every 50 percent increase in Anthropic’s value above a baseline that has not yet been disclosed, the board can add another 2 percent of the shares. At the valuation of more than $2 trillion that Anthropic’s backers are hoping for in its planned I.P.O., 5 percent would be worth about $100 billion.
The money could go toward large donations to nonprofits or direct payments to specific groups of people. Anthropic explicitly names the impact of A.I. on the labor market as an example, an area where the company itself is already leaving a mark: Junior jobs are shrinking at Anthropic. Anthropic has already committed $200 million to research on economic futures but says “bolder approaches” may be needed. The fund may not use its shares for lobbying or political donations.
Together with earlier commitments, Anthropic says, about 20 percent of its shares are now dedicated to the public good. Those include the co-founders’ pledge to donate 80 percent of their equity and a program in which the company matches employees’ share donations. Anthropic plans to share details on governance and valuation milestones in the coming months.
The competition responded with a jab. “I recommend giving the shares directly to the public,” Elon Musk, the head of xAI, wrote on X, adding: “This is what I will be doing.” He did not say how or when.
Intervention Only Above 50 Deaths
Anthropic is less generous when it comes to regulation. According to Politico, the company has presented congressional staff with a draft A.I. safety bill. It would apply only to developers of frontier models trained with more than 10²⁵ computing operations that also either generate more than $500 million in annual A.I. revenue or spend more than $1 billion on A.I. research. These companies would have to publish safety frameworks and transparency reports, allow independent evaluations, report incidents to the Commerce Department within 72 hours and protect whistle-blowers.
Only the Commerce Department could stop a model, however, and only by going through a federal court. That would require a foreseeable, imminent risk of more than $1 billion in damage or of more than 50 people being killed or seriously injured. The draft names three dangers: helping to create biological weapons, large-scale cyberattacks and models that evade their developers’ safety controls. No other federal agency could independently pull a model.
Clear Rules or Too High a Bar?
Anthropic defends the draft. A spokesperson said it reflects the proposals in the company’s “Advanced AI Framework” and gives the federal government the ability “to block or deter dangerous deployments of the most powerful models.” A single federal standard would also replace the patchwork of state laws.
Critics see it differently. The draft would effectively make the Commerce Department the only agency able to stop an A.I. model, and only once dozens of lives are at stake, Tom’s Hardware writes. Others argue that compliance costs favor established labs like Anthropic over smaller rivals. The draft has not yet been formally introduced.
