The United States government has lifted foreign access restrictions on AI models Fable 5 and Mythos 5, developed by Anthropic. This decision follows closed-door negotiations between the startup and the Trump administration regarding safety concerns. While the specific terms of the agreement remain undisclosed, the reversal follows Anthropic's implementation of new safeguards.

The initial ban was prompted after Amazon alerted the US administration that existing safeguards on the Fable model could be bypassed. To address this, Anthropic introduced classifiers, which are smaller AI systems designed to detect and block requests related to prohibited cybersecurity activities. The Commerce Department’s Center for AI Standards and Innovation (CAISI) supported this technical fix. Despite these changes, the lack of transparency surrounding the agreements has raised concerns about the predictability and openness of US AI policy for international users.

The opacity regarding how these safety standards were negotiated creates uncertainty for those relying on American AI infrastructure. Critics argue that clear communication is necessary to maintain trust in these tools, as the undisclosed nature of the compliance measures leaves developers and users without insight into the specific safety protocols now in place for Fable 5 and Mythos 5.

The United States government has lifted foreign access restrictions on AI models Fable 5 and Mythos 5, developed by Anthropic. This decision follows closed-door negotiations between the startup and the Trump administration regarding safety concerns. While the specific terms of the agreement remain undisclosed, the reversal follows Anthropic's implementation of new safeguards.

The initial ban was prompted after Amazon alerted the US administration that existing safeguards on the Fable model could be bypassed. To address this, Anthropic introduced classifiers, which are smaller AI systems designed to detect and block requests related to prohibited cybersecurity activities. The Commerce Department’s Center for AI Standards and Innovation (CAISI) supported this technical fix. Despite these changes, the lack of transparency surrounding the agreements has raised concerns about the predictability and openness of US AI policy for international users.

The opacity regarding how these safety standards were negotiated creates uncertainty for those relying on American AI infrastructure. Critics argue that clear communication is necessary to maintain trust in these tools, as the undisclosed nature of the compliance measures leaves developers and users without insight into the specific safety protocols now in place for Fable 5 and Mythos 5.