Anthropic Pre-IPO Shares Drop 3.7% After US Orders Major AI Model Shut

• Anthropic pre-IPO shares fell 3.7% on Hyperliquid after the US government ordered a shutdown of Fable 5 and Mythos 5.
• Anthropic pre-IPO shares had been trading near all-time highs above $1,800 in the days following Fable 5’s launch.
• The government cited a potential jailbreak vulnerability, but Anthropic says the same flaw exists in OpenAI’s GPT-5.5 without any bypass.
• Anthropic argues that applying the government’s standard across the industry would effectively halt all frontier model deployments.

• Anthropic pre-IPO shares fell 3.7% on Hyperliquid after the US government ordered a shutdown of Fable 5 and Mythos 5.
• Anthropic pre-IPO shares had been trading near all-time highs above $1,800 in the days following Fable 5’s launch.
• The government cited a potential jailbreak vulnerability, but Anthropic says the same flaw exists in OpenAI’s GPT-5.5 without any bypass.
• Anthropic argues that applying the government’s standard across the industry would effectively halt all frontier model deployments.

Anthropic Pre-IPO Shares Take a Hit as Washington Steps In

Anthropic pre-IPO shares dropped sharply over the weekend after the US government ordered the company to shut down its two most powerful AI models — a move that spooked traders and raised uncomfortable questions about the company’s path to a public listing. The Anthropic perpetual contract on Hyperliquid, a cash-settled derivative that functions as a proxy for the company’s expected IPO price, fell roughly 3.7% on Saturday to around $1,627. That’s a meaningful slide from the highs above $1,800 the contract had been touching just days earlier, in the immediate glow of Fable 5’s launch.

The directive came late Friday. Anthropic says it received the shutdown order at 5:21 p.m. ET, issued under national security authorities, requiring the company to suspend access to both Fable 5 and Mythos 5 for any foreign national — regardless of whether they’re inside or outside the United States. In practice, Anthropic went further and pulled both models for all customers entirely. The company complied. But it didn’t stay quiet about it.

What the Government Actually Found — and Why Anthropic Is Pushing Back

The trigger for the shutdown was a reported jailbreak technique targeting Fable 5. The government told Anthropic it had discovered a method to bypass the model’s safety guardrails. Anthropic reviewed what it was shown and came back with a pointed response: what the government described isn’t a universal jailbreak. According to the company, the technique exploits a narrow set of previously known, minor vulnerabilities — and crucially, those same vulnerabilities can be surfaced by OpenAI’s GPT-5.5 without any bypass technique at all.

That’s a significant counter-argument. Anthropic is essentially saying: if this is the bar for a forced shutdown, then every frontier AI lab should be offline right now. The company described the alleged exploit as something close to asking a model to read a codebase and flag software flaws — a routine defensive security task performed by developers and researchers every single day. Applying the government’s apparent standard to the whole industry, Anthropic warned, “would essentially halt all new model deployments for all frontier model providers.”

The company also noted that the government has, so far, provided only verbal evidence of the vulnerability — no written documentation, no technical report, no formal disclosure through established channels. That’s a striking detail. For a company that has built its entire identity around responsible AI development and transparent safety practices, being told to switch off its most capable models based on oral briefings alone is a hard pill to swallow — and they’re saying so out loud.

The IPO Overhang Is Real

With open interest on the Hyperliquid perpetual sitting near $8.6 million, the Anthropic contract is small compared to something like the SpaceX perp. But in the context of a company that hasn’t yet filed for an IPO, it’s a meaningful barometer of where sophisticated traders are pricing future equity. Anthropic pre-IPO shares, as reflected by the perpetual contract, are moving in the wrong direction right now.

The timing is particularly awkward. Fable 5 launched to considerable industry fanfare, and the contract’s surge past $1,800 reflected genuine optimism about Anthropic’s competitive position — and the possibility that a public offering could be closer than previously thought. That narrative took a direct hit when Washington intervened. Anthropic pre-IPO shares sliding back to $1,627 isn’t catastrophic, but it reflects a real recalibration of risk. Investors in pre-IPO markets don’t love regulatory uncertainty at the best of times. A forced model shutdown tied to national security concerns — especially one the company is publicly contesting — is exactly the kind of headline that makes institutional investors more cautious.

The Broader Stakes for AI Regulation

This situation puts a spotlight on a tension that’s been simmering in the AI industry for some time: how does a government exercise national security oversight over AI systems without applying standards that are either technically illiterate or selectively enforced? Anthropic’s core argument — that the identified vulnerability exists in competitor models too — is a challenge to the coherence of the directive itself, not just its application.

There’s also a competitive dimension worth considering. If Fable 5 and Mythos 5 remain offline while GPT-5.5 continues operating normally, Anthropic loses ground in one of the most aggressively contested markets in tech. Every day those models are unavailable is a day enterprise customers are being pushed toward alternatives. The company has worked hard to differentiate itself on safety — the idea that its safety-first approach could be weaponised against it by regulators is an irony that won’t be lost on anyone watching from the outside. The pressure on Anthropic pre-IPO shares will only intensify if the shutdown drags on.

Anthropic has said it will publish a detailed technical rebuttal within 24 hours of the shutdown, laying out specifically what the government showed them and why they believe it doesn’t meet the threshold for a forced suspension. That publication will be important for several reasons. It’ll give independent security researchers something to evaluate. It’ll frame the public debate before the government responds. And it could form the basis of any legal challenge if the order is extended or not reversed.

What Happens Next for Anthropic’s IPO Ambitions

The immediate question is straightforward: does the government reverse the order, narrow it, or expand it once Anthropic’s technical rebuttal is on the table? If the shutdown is lifted quickly — particularly if Anthropic’s argument that other models carry the same risk lands with regulators — Anthropic pre-IPO shares will likely recover and the episode becomes a footnote. If the order stands, or gets broader, the IPO calculus changes significantly.

Longer term, this is a preview of a genuinely difficult regulatory environment. Governments around the world are scrambling to build frameworks for AI oversight, and the US has been navigating a complicated balance between promoting domestic AI leadership and guarding against national security risks. Anthropic is now in the uncomfortable position of being the test case for where those two priorities collide. For anyone tracking Anthropic pre-IPO shares as a leading indicator of the company’s public market prospects, how this particular dispute resolves — and what precedent it sets for how national security claims translate into forced model shutdowns — will matter well beyond one company’s IPO timeline.

Source: CoinDesk

Frequently Asked Questions