The U.S. government is reportedly unlikely to extend Anthropic-related AI export controls to other companies.
Security Concerns Drive Selective Regulatory Action
According to a report by The Information on Saturday, citing a source, the administration’s concerns remain focused on security vulnerabilities identified in Anthropic’s Fable 5 and Mythos 5 models.
According to the report, the source said the government was concerned that Fable could still “find vulnerabilities in sensitive systems” despite guardrails designed for broader public release. The administration also cited risks that advanced AI capabilities could spread to foreign adversaries.
The two “Mythos-class” AI models, built for highly complex reasoning and long-horizon agentic work, were disabled following a U.S. government order restricting access to foreign nationals.
Regulatory Pressure Builds
On Friday, in a blog post, Anthropic argued that similar vulnerabilities exist in other publicly available models, including recent releases from OpenAI. The AI startup added, “The potential jailbreaks that have been disclosed to us are either entirely benign responses or are minor findings.”
Reports have also circulated that Amazon.con Inc. (NASDAQ:AMZN) CEO Andy Jassy previously raised security concerns tied to Anthropic’s models, which contributed to earlier regulatory scrutiny.
Talking about the matter on X, former White House AI and crypto czar David Sacks, who currently co-chairs the President’s Council of Advisors on Science and Technology, said Anthropic CEO Dario Amodei was asked to fix the vulnerabilities or de-deploy the model but declined.
The episode underscores increasing regulatory scrutiny of frontier AI systems amid rising focus on export controls and cybersecurity standards in AI model deployment.
Photo Courtesy: Shutterstock
Disclaimer: This content was partially produced with the help of AI tools and was reviewed and published by Benzinga editors.
Recent Comments