Dozens of cybersecurity experts have petitioned the White House to lift export restrictions on Anthropic's Fable and Mythos models, warning that the controls hamper defensive capabilities against digital threats.
The group contends that limiting access to these advanced AI models undermines cybersecurity professionals who rely on cutting-edge tools to identify and patch vulnerabilities in software and systems. Export controls, they argue, create a two-tier system where US defenders operate at a disadvantage while adversaries may access comparable technology through other channels.
Anthropic's Fable and Mythos represent the company's most capable models to date. The US government imposed restrictions citing national security concerns about advanced AI technology reaching hostile actors or being used for malicious purposes. The move reflects broader scrutiny of frontier AI exports, particularly models that could assist in cyberattacks or weapons development.
The cybersecurity community's protest highlights a genuine tension in AI policy. Defensive applications such as vulnerability discovery, threat analysis, and security testing often require the same capabilities that could theoretically enable attacks. Restricting access to powerful models may protect against worst-case scenarios but simultaneously weakens legitimate defenders who operate within legal and ethical frameworks.
This dispute echoes longstanding debates over encryption and security tools. Export controls on powerful cryptography were loosened decades ago partly because restrictions harmed American companies and defenders more than adversaries. The cybersecurity experts essentially argue history should repeat itself with AI models.
The petition carries weight given the signatories' credentials in defensive security. Their concern addresses a practical problem: if US organizations cannot access state-of-the-art AI for security work, they fall behind in the ongoing arms race against increasingly sophisticated cyber threats.
The White House has not yet responded to the petition. The outcome could shape how the US government approaches future AI export restrictions and whether it prioritizes absolute containment or accepts calculated risks to maintain defender