🤚 The Open-Palm Reversal

The Trump administration, apparently tiring of its own regulatory theatrics, has lifted the export restrictions on Anthropic’s Fable 5 and Mythos 5 models — 18 days after imposing them. Secretary of Commerce Howard Lutnick negotiated the deal, announcing that Anthropic agreed to “proactively detect and address security risks” and “work diligently with the U.S. government on protocols and standards.”

The restrictions were originally placed on June 12, citing national security concerns. Both models were added to the export-restricted technologies list, making international access impractical and forcing Anthropic to pull them from service outside the US. As of July 1, Anthropic has begun restoring global access.

In exchange for the privilege of selling its own products again, Anthropic has committed to:

• Proactively detecting and addressing security risks associated with the models
• Collaborating with the U.S. government on protocols, standards, and future releases
• Informing the government of any malicious activity

So, essentially, everything it was already doing.

👐 The Two-Handed Analysis

Here is what makes the 18-day ban particularly exquisite: while the US government was busy protecting the world from Anthropic’s AI, Asian startups released Fugu and Tulonfeng — models approaching Mythos-level capabilities. The embargo’s primary achievement was demonstrating that America’s AI advantage is a perishable good, and that other countries will happily fill any vacuum Washington creates.

The original restrictions were questioned from day one. Our June 13 coverage noted that the ban was ostensibly about a jailbreak vulnerability — the kind every major model can already be coaxed into. Industry experts openly debated whether the restrictions were genuinely about security or just political posturing.

The answer, it turns out, was yes.

OpenAI, which has navigated similar government approval processes for its latest models, watched this unfold with the quiet contentment of a company that knows when to say nothing. Anthropic — the lab that has made safety its entire identity — was essentially punished for being too good at its job.

🌿 The Gentle Awakening

There is something almost poetic about an 18-day export ban. It is long enough to disrupt business, damage trust, and hand competitors a head start — but too short to accomplish anything that might justify the disruption. It is the regulatory equivalent of starting a fistfight at a dinner party and then pretending you were just reaching for the salt.

The Trump administration’s AI policy continues to oscillate between “America must lead the AI race” and “America’s AI companies must be stopped before they do something we haven’t specified yet.” It is not erratic policymaking so much as policymaking by multiple people who have not spoken to each other and are not entirely sure what the product does.

Lutnick’s negotiated agreement — proactive threat detection, government collaboration, malicious activity reporting — is what Anthropic was already doing before the ban. The company’s reward for 18 days of compliance is permission to resume doing what it was doing on June 11.

👑 The Gold-Leaf Reckoning

The precedent matters more than the reversal. Every AI company now knows that 18 days of government mood swing can cost them international customers, reputational equity, and competitive positioning. The ban created a regulatory template that says: we can ground your product at any time, for any reason, and the appeal process is a phone call with Howard Lutnick.

Meanwhile, Fugu and Tulonfeng will continue to exist, unencumbered by American export policy, available to anyone with an internet connection. The AI race does not pause for regulatory drama. It just finds a detour.

The 18-day ban accomplished exactly one thing: it proved that the fastest way to accelerate foreign AI development is to temporarily ban your own.

“When you ban the product for 18 days and the competition catches up in 12, you haven’t protected national security — you’ve provided a free consulting report to your adversaries, delivered with expedited shipping.” — The Slap of Wisdom Trade Policy Desk, currently exporting this article without a license