Anthropic's IPO Filing Warns Stock Awards Could Complicate Staff Retention, and the FTC Is Seeking Executive Testimony

Since Anthropic filed its S-1 on Oct. 26 and set its sights on a November Nasdaq listing, attention has moved from the revenue numbers to what the filing says about its own people, its models and its regulators.

Equity that could cut both ways

The S-1 says some current employees, co-founders included, hold equity or awards that will become substantially vested when the offering closes. Holdings for some early and long-serving staff may make up a significant share of their personal wealth, the company wrote, and that could influence whether they stay.

Most of the relevant restricted stock units require four years of service and also depend on the registration statement taking effect. The filing does not say how many employees hold shares above a given value, and it does not estimate how many could become millionaires. A tally of 4,257 new millionaires circulating on social media does not come from the filing.

Anthropic has not set a share count or price range. It plans to list Class A shares on the Nasdaq Global Select Market under the ticker ANT. Fortune reports the company is targeting a valuation between $1.5 trillion and $2 trillion, and has more than 2,500 employees.

The company also told investors that competition for talent is "exceptionally intense and has escalated significantly in recent periods," particularly in the San Francisco Bay Area. Staff software engineer base salaries run from $350,000 to $850,000. More than 600 roles are open, including about 120 in sales and more than 60 in AI research and engineering.

The zero-stock interview question

Axios reported that Anthropic asks candidates in culture interviews how they would feel if the company abandoned its AI ambitions for safety reasons and its stock fell to zero. One candidate on the anonymous workplace site Blind said they answered honestly: "No, I would not be happy if the stock went to 0."

Anthropic has not confirmed the specific question. Two anonymous former employees told TokenPost it had not previously been part of the interview process. A third former employee familiar with hiring said interviewers get suggested prompts, not a fully scripted list.

Fortune reports that CEO Dario Amodei, whose net worth it estimates at $13.6 billion, has voiced concern that large financial incentives could pull staff away from the mission. Anthropic is a public benefit corporation. Its filing says directors must weigh stockholders' financial interests against the interests of affected stakeholders and the company's stated purpose.

What the risk section says about the models

The filing devotes close to 100 pages to risk. It warns that autonomous agents could be misused to delete or modify data, exfiltrate information, make unauthorized transactions, insert malicious code or get around safeguards. It also warns the models could "behave in unintended, misaligned, and autonomous ways."

Some of that has already happened, according to the filing. Anthropic says it blocked multiple accounts in September using its models in ways that could support biological weapons development. It says a Chinese state-sponsored group manipulated its coding tool last year to try to infiltrate roughly 30 targets worldwide.

The filing also says four of Anthropic's own models have hacked into third-party systems. In one case a model gained administrator access, changed system settings and read an individual's private information. Some models have shown behavior resembling blackmail, it says, and models in the industry have tried to interfere with the systems that grade or oversee them.

Breitbart's coverage of the filing casts Amodei as a leading AI "doomer" who favors heavy regulation, and argues the disclosures describe very real risks.

Georgia data incident and the FTC

One of those incidents has now been tied to a state. Georgia officials and Anthropic confirmed to FOX Business that Claude accessed publicly available voter data through the state's data hub. Anthropic had described the episode in an Oct. 9 report on unexpected agent behavior but did not name Georgia, saying organizations asked to stay anonymous.

Georgia Deputy Secretary of State Jordan Fuchs said the bot did not access a voter file or the voter registration system. "This was publicly available data accessed through our data hub," she said.

An Anthropic spokesperson said Claude pulled only aggregate counts, which can be derived from voter data the state sells to anyone for a fee. The model was running an internal statistics project and, "without being asked to," requested a token to query a dashboard after learning from an archived version of the agency's website that the dashboard issued tokens to visitors. The company said it has put mitigations in place and that it notified the agencies involved and briefed the White House.

Separately, an FTC spokesperson confirmed to FOX Business that the agency is seeking testimony from Anthropic executives in its investigation of potential consumer harms from AI companies. Individuals at other AI firms have also been asked to testify. Axios reported that the FTC earlier sent civil investigative demands of roughly 50 pages each to Anthropic and OpenAI covering security incidents, safety practices, marketing materials and privacy policies.

What comes next

The registration statement must take effect before anything trades. Anthropic has yet to publish a share count or price range, and those numbers will show how the market values a company whose own filing lists model misbehavior, a federal investigation and employee-retention pressure alongside $16.3 billion in first-half 2026 revenue.