The two most valuable private AI companies in the world just told a chunk of their would-be investors: you don’t actually own what you think you own.

Anthropic and OpenAI both moved this week to declare unauthorized secondary sales of their private shares invalid, targeting the Special Purpose Vehicle structures that have become the go-to mechanism for retail and smaller institutional investors trying to get exposure to pre-IPO AI companies. Anthropic went further, specifically naming Forge Global among the platforms facilitating trades it considers illegitimate.

What actually happened

Anthropic declared that any stock sales conducted without its formal approval are “void.” Not questionable, not under review. Void. The company stated it will not recognize buyers in these transactions as shareholders, effectively rendering their purchased stakes worthless pieces of paper.

OpenAI took a nearly identical position, stating that any share transfer without written consent is unauthorized and invalid.

The core issue centers on SPVs, which are essentially shell entities created to pool investor capital and buy shares in private companies. Think of them as a carpool lane for people who can’t afford the solo ride. An SPV buys a block of shares, then sells fractional interests to multiple investors who each get a slice of the upside.

In English: instead of needing millions to buy shares directly from an employee or early investor, you could throw in a smaller amount through an SPV and technically hold exposure to Anthropic or OpenAI equity. The problem is that both companies explicitly prohibit this kind of transfer without their blessing. And they’ve decided to stop being polite about enforcement.

Anthropic specifically called out SPV structures as a prohibited method of acquiring its shares, a move that puts platforms facilitating these transactions on notice. By naming Forge Global directly, Anthropic signaled it’s not issuing vague warnings. It’s pointing fingers.

Why crypto markets should care

Here’s the thing. This isn’t just a Silicon Valley corporate governance story. The fallout extends directly into crypto, where a growing number of protocols and platforms have built products offering tokenized or synthetic exposure to private equity stakes in companies like Anthropic and OpenAI.

The logic behind these products is straightforward: wrap a claim on private company shares into a token, make it tradeable on-chain, and let crypto-native investors access an asset class that was historically reserved for venture capitalists and accredited investors. The demand has been enormous, driven by the AI hype cycle and the perception that getting in before an IPO is practically a guaranteed windfall.

But if the underlying shares are deemed void by the issuing company, the token representing those shares is, by extension, a claim on nothing. It’s like having a deed to a house that the county doesn’t recognize. You can trade the deed all you want, but nobody’s letting you through the front door.

The enforceability question is now front and center. Any tokenized product, on-chain wrapper, or synthetic derivative that derives its value from unauthorized Anthropic or OpenAI shares has a fundamental legal problem. The issuer of the underlying asset says you’re not a shareholder. Good luck arguing otherwise in court.

This creates a two-tier risk structure. Investors who bought tokenized exposure face not only the standard volatility risk of private equity, but now a binary legal risk: the possibility that their entire position could be worth zero if the company enforces its transfer restrictions. And this week, both companies made clear they intend to do exactly that.

The bigger picture for pre-IPO investing

The crackdown reflects a broader tightening across the private equity landscape. As investor demand for exposure to high-profile private companies has surged, so has the cottage industry of secondary-market intermediaries promising access. Some of these platforms operate within the rules, securing proper transfer approvals. Others have treated company consent as more of a suggestion than a requirement.

For years, private companies largely looked the other way on unauthorized secondary trades. The amounts were small, the frequency was low, and pursuing enforcement was more hassle than it was worth. That calculus has changed.

Anthropic and OpenAI are now among the most valuable private companies on the planet. The stakes are higher. The volume of unauthorized secondary activity has grown. And both companies have strategic reasons to maintain tight control over their cap tables, particularly as they navigate potential future IPOs, government scrutiny, and complex relationships with investors like Google, Microsoft, and Amazon.

Maintaining cap table discipline matters for IPO readiness. A messy shareholder registry with hundreds of unauthorized SPV investors creates legal and regulatory headaches that no company wants to deal with during an already complicated public offering process.

For crypto platforms that have built products around pre-IPO share exposure, this week’s announcements should be a wake-up call. The legal foundation of these products was always somewhat shaky. Now it’s been explicitly challenged by the issuers themselves.

Investors holding tokenized Anthropic or OpenAI shares should be asking their platform providers one very specific question: did the issuing company approve the underlying share transfer? If the answer is anything other than an unqualified yes, the investment thesis has a hole in it that no amount of blockchain transparency can fix.

The irony is rich. Crypto was supposed to disintermediate gatekeepers and democratize access to assets. But when the asset issuer can simply declare your ownership void, decentralization doesn’t help much. Some gates, it turns out, still have keepers.