After 15 days of standoff, the Trump administration has partially lifted its historic export ban on Anthropic’s most powerful AI model — but the questions swirling around why the ban happened in the first place haven’t gone away.

Story Highlights

• The Trump administration issued the first-ever U.S. export ban on an AI model on June 12, 2026, targeting Anthropic’s Claude Fable 5 and Mythos 5 over national security concerns.
• Commerce Secretary Howard Lutnick sent a directive to Anthropic CEO Dario Amodei requiring a license for any export, re-export, or domestic transfer of the models.
• The ban was partially lifted after 15 days, allowing Mythos 5 to be used by a small group of trusted cybersecurity partners.
• Critics, including more than 100 cybersecurity experts, called the ban an overreaction — and questions remain about whether the process was fair or politically motivated.

America’s First AI Export Ban

On June 12, 2026, the Trump administration made history by placing the first-ever U.S. export ban on an artificial intelligence model. Commerce Secretary Howard Lutnick sent a directive to Anthropic CEO Dario Amodei, requiring a government license before the company could export, re-export, or transfer its Claude Fable 5 and Mythos 5 models. The administration classified Mythos as a national security asset, citing what it called exceptional hacking capabilities built into the model. ^[5]

White House adviser David Sacks said a trusted testing partner had found a jailbreak — a way to bypass the model’s safety guardrails — that could unlock Mythos-level cyber capabilities. Sacks confirmed the administration had first asked Anthropic to fix the flaw or pull the model before resorting to the export order. The ban also extended to foreign nationals working inside the United States, including Anthropic’s own employees. ^[1]

Anthropic Pushes Back Hard

Anthropic said it received the directive without being given specific details about the national security concern. CEO Dario Amodei went further, arguing the government had pre-decided to impose export controls regardless of the evidence. The company was given only 90 minutes to comply — a window critics say was too short for any real technical review. ^[8]

The government also claimed Amodei was unreachable during the crisis because he was on a wellness retreat. Journalist Ashley Vance disputed that claim directly, saying Amodei was present at Anthropic at the time. That contradiction raised eyebrows and added fuel to the debate over whether the administration handled this fairly. ^[10]

Was the Threat Real or Overblown?

An independent cybersecurity firm reviewed the government’s report and told the Wall Street Journal the alleged jailbreak involved Mythos helping to patch security vulnerabilities — a defensive action, not an attack. The firm called the government’s response an overreaction. More than 100 cybersecurity leaders, including Alex Stamos, signed a public letter urging Washington to lift the ban, warning it weakens the very people defending U.S. networks.

The US government just partially lifted its ban on Anthropic's Claude Mythos 5.

Here's the wild backstory:

🔴 June 12: Trump admin hits Anthropic with an export control order, Mythos 5 & Fable 5 shut off for ALL users, including Anthropic's own foreign employees.

🟡 Trigger:…

— Ankit Sachan (@iankitxai) June 27, 2026

The ban also strained ties with U.S. allies. Because the order applied equally to allied nations, countries in Europe and elsewhere found themselves locked out of a tool many of their cybersecurity teams relied on. Reports from South Korea noted the ban could push countries to build their own AI models faster, reducing their dependence on U.S. technology — the opposite of what American AI strategy aims to achieve. ^[2]

Political Favoritism Clouds the Picture

Some observers pointed to a troubling pattern. Reports noted that Anthropic was excluded from Trump administration equity talks involving AI companies, while an OpenAI-linked political action committee called “Leading the Future” spent $25 million supporting Trump. Former Pentagon official Pete Hegseth publicly disparaged Anthropic on social media. These facts don’t prove the ban was political — but they raise fair questions that the administration has not fully answered.

After 15 days of negotiations, the government cleared Anthropic to redeploy Mythos 5 to a limited group of trusted cybersecurity defense partners. That partial lift suggests the administration recognized the ban was too broad. Protecting genuinely dangerous AI capabilities from foreign adversaries is a legitimate goal — one conservatives broadly support. But the rushed process, the lack of transparency, and the unanswered questions about political motives mean this story isn’t over. The full technical report from the government’s testing partner has still not been made public.

Sources:

[1] Web – Anthropic’s AI Export Ban on Claude Mythos Tool Partially Lifted by …

[2] Web – US export control on Anthropic’s Claude Fable 5 and Mythos 5

[5] Web – White House Imposes Export Controls on Anthropic’s Mythos AI Model

[8] Web – US Imposes Export Controls on Anthropic Mythos AI Models – LinkedIn

[10] Web – US bans Anthropic from exporting most advanced AI systems to other …