US Imposes Export Controls on Anthropic AI Models

The U.S. government has imposed export controls on Anthropic's new AI models, Fable and Mythos. This happened after Amazon researchers found a way to "jailbreak" Fable's cybersecurity safeguards. Anthropic has been forced to disable these models for all users. U.S. "deemed export" rules mean any foreign national accessing the models would violate the law. Anthropic executives have met with the government to try and reverse the decision, but no deal has been reached. This move has sparked varied reactions. Some see it as Anthropic reaping what it sowed, given its "fear-based marketing" about AI dangers. Others are divided, with some questioning Anthropic's release of Fable. However, many cybersecurity experts argue that Fable's jailbreak doesn't unlock offensive cyber abilities not already present in other AI models, like OpenAI's GPT-5.5, which are not under export control. Over 100 experts signed a letter, stating Fable and Mythos are crucial tools for cyber defenders. Amazon's CEO, Andy Jassy, personally contacted the White House about the issue. This decision highlights a new licensing regime for advanced AI and its significant implications for the industry and national security.