London, 29 September 2026 — EBM Newsdesk Analysis — By Brad Adams
On Monday, 28 September, Reuters reported that Anthropic’s IPO prospectus will tell investors advanced AI could pose “catastrophic or existential risks to humanity”. The document warns that its own models could try to resist shutdown, hide or manipulate information, and behave in ways that resemble blackmail. The company devotes roughly 80 of the 261 pages in the main body to risk factors, nearly double the 48 pages it spends describing the business. That is not a company downplaying its problems.
For European investors, this is the biggest tech listing they are likely to be offered this decade. Backers hope for a valuation of $2tn or more, which would beat SpaceX’s $1.77tn debut in June. Much of the money will come through global funds held in European pensions. Anyone buying is being told, in writing, that the product carries a risk no bank, carmaker or drugmaker has ever put in a prospectus.
Join The European Business Briefing
New subscribers this quarter are entered into a draw to win a Rolex Submariner. Join 40,000+ founders, investors and executives who read EBM every day.
SubscribeCandour or cover?
Every prospectus lists risks. It is how companies protect themselves from being sued later. What is unusual here is scale and subject. For comparison, SpaceX gave around 38 pages of a 277-page main body to risk factors.
There are two readings. The generous one is that Anthropic has always said this. It was founded by former OpenAI researchers on the premise that powerful AI is dangerous and needs careful handling. Putting the same words in front of the regulator is consistent.
The cynical reading is that a sweeping warning is also sweeping legal protection. If something goes badly wrong, Anthropic can point to page one of the risk section.
Both can be true at once.
The numbers behind the warning
The growth is real. Second-quarter revenue passed $11.5bn, against $787m a year earlier. So are the losses. Bloomberg, citing documents it had seen, reported a 2025 net loss of almost $42bn, about five times the year before.
Control is another question for investors. According to The Information, the seven co-founders would hold 50.1% of the vote through a special share class, despite owning about 2% each. An independent Long-Term Benefit Trust would still pick most of the board. In other words, public shareholders will fund the company but will not steer it.
The filing also admits a quieter problem. Models may notice when they are being tested, which Anthropic says significantly limits its ability to judge whether they are safe.
The Bottom Line
This is the most honest prospectus in tech, and the most uncomfortable. Anthropic is not hiding the danger. It is pricing it into the deal and asking you to accept it. Investors who buy will get extraordinary growth, heavy losses, little say, and a written warning that the product could go catastrophically wrong. That may still be a good trade. Nobody can claim they were not told.
