A coalition of cybersecurity leaders is calling on the US government to lift export restrictions on Anthropic's security-focused AI models. They argue the current controls are undermining America's ability to defend its networks and could inadvertently strengthen open-source competitors abroad.
Why the Controls Are Under Fire
The restrictions, imposed under national security rules, limit who can access Anthropic’s most advanced defensive AI tools. Cyber leaders say this creates a gap: US agencies and critical infrastructure operators can't deploy the models as widely as needed while adversaries face no such barriers. The result, they warn, is a net loss for American cybersecurity.
Anthropic builds AI systems designed to detect and neutralize threats faster than traditional tools. But export rules classify those models as sensitive technology, making it hard to share them even with trusted allies or domestic partners.
The Open-Source Risk
One unintended consequence the cyber leaders highlighted: the export controls could push the market toward open-source alternatives. If US companies can't access Anthropic's models, developers in other countries may build their own—and release them freely. That would dilute the competitive edge of American AI security tools while giving hostile actors a peek at the technology.
“You're basically telling the rest of the world to build their own defensive AI, and they will,” one former government official familiar with the discussions said. “That doesn't help us.”
What the Cyber Leaders Want
The group hasn't asked for a blanket rollback. Instead, they're pushing for a targeted exemption: let Anthropic’s security models flow to vetted partners—especially critical infrastructure firms, hospitals, and defense contractors—without a license. They argue the current system treats a defensive capability the same as an offensive weapon.
The ask lands as Washington debates broader AI export policy. The Commerce Department's Bureau of Industry and Security oversees the controls and has shown no sign of moving quickly. A review could take months.
What Happens Next
No public hearing has been scheduled. The cyber leaders are now trying to build a coalition of industry and former national security officials to press the case. Whether the White House sees value in loosening the rules—or views the open-source risk as exaggerated—remains the open question.
