Anthropic Asked for Regulation, Now Calls It Risk
Anthropic's IPO prospectus warns that Washington's stance toward the company could scare off commercial customers, a notable disclosure from the AI lab that has spent the past year advocating for government AI oversight.
Anthropic’s IPO prospectus warns that Washington’s attitude could damage ties with commercial customers, creating business risks that extend beyond federal contracts, per Reuters.
In the filing, Anthropic tells investors that Washington’s perception of the company, not just its policy decisions, is itself a variable that could cost it business.
While government agency contracts account for under 1 percent of annual revenue, the prominence of the warning makes it a notable disclosure rather than routine contract dependency.
A Suspension That Already Happened, Not a Hypothetical
The prospectus didn’t rely on abstract risk language; it pointed to a real event.
In June, the Commerce Department imposed worldwide export restrictions on Anthropic’s two most capable models, Fable 5 and Mythos 5, citing concerns over a technique that could let the models read code and fix software vulnerabilities.
Anthropic disabled both models for every customer globally to ensure compliance, not just the users the restriction targeted, a decision that cost the company revenue and access for customers who had nothing to do with the underlying concern.
Though the restriction was lifted on July 1, the episode serves a dual purpose in the filing: proving rapid compliance to regulators while demonstrating to investors how a single government decision can take flagship products offline overnight.
The FTC Probe Arrives at the Worst Possible Moment
Reuters reported Wednesday that the FTC is conducting an industry-wide investigation into AI companies including Anthropic, a probe landing just as the company tries to market what could be a $2 trillion valuation to public investors.
Amodei met with Trump for dinner last Sunday as calls for stronger AI regulation have grown louder, though Trump has largely rejected those calls.
Anthropic devoted roughly 80 of its prospectus’s 261 pages to risk factors, including model self-preservation and blackmail behaviors, signaling a deliberate push to address scrutiny before public markets do.
The Company Asking for Rules Is Now Afraid of Them
Anthropic has spent the past year as the AI industry’s loudest voice for more oversight, Amodei pushing for coordinated safety pacing, the company backing California’s SB 53 safety-disclosure law, and publicly advocating strict export controls on AI chips.
Yet, the company simultaneously contested the June export restrictions on its own models.
This creates a delicate tension: advocating for broad regulatory boundaries while identifying those same interventions as key revenue risks in its prospectus.
The temporary suspension of its Fable and Mythos models highlights how safety commitments and commercial realities do not always align, a trade-off Anthropic must now explicitly outline for potential investors.
Sources:
