Frontier artificial intelligence (“AI”) companies are beginning to tap the public markets through some of the largest initial public offerings (“IPO”) in history. Anthropic is the first of these to approach the market, and this memo uses it as a case study for the questions that any frontier AI IPO should raise. Anthropic will not be the last. OpenAI is widely reported to be preparing its own public offering, and a broader wave of frontier AI listings is likely to follow. The questions Anthropic’s filing raises—about early offloading of shares by insiders, circular financing, off-balance-sheet leverage, legal exposure, and “catastrophic risks”—will recur in each of them, so the disclosures made here will shape every AI IPO that comes after. 

The Securities and Exchange Commission’s (“SEC” or “Commission”) core mandate in reviewing an IPO is the protection of investors through full and fair disclosure. That mandate is unusually consequential here. An offering of this size is pulled into the broad market indexes within days of listing, and from there into the index funds that form the backbone of ordinary Americans’ 401(k)s and pension plans. The quality of Anthropic’s disclosures is therefore of relevance to nearly everyone with a retirement account, not only the sophisticated private investors who can evaluate the risks and rewards for themselves. If Anthropic fails—if its complex web of financing comes undone or its systems cannot be deployed safely—the losses will be borne not just by insiders, but by the millions of ordinary Americans whose retirement savings may collapse or potentially be extinguished.

This memo proceeds in four parts. Section I sets out the SEC’s disclosure mandate and the tools available to it and to other fiduciaries during the IPO process. Section II turns to what third parties can do. Section III provides the factual background on the various types of risks (from the financial to the existential) and potential legal liabilities that make frontier AI offerings unusual. Section IV sets out the questions the SEC and those fiduciaries should ask, first of any frontier AI company, and then of Anthropic specifically. Even if these measures do not ultimately stop an IPO, they serve a vital purpose: equipping investors to make informed decisions and forcing risk to be priced more accurately, so that valuations reflect genuine risk.