The US government has backed away from export restrictions on Anthropic’s Fable and Mythos AI models, marking a major reversal less than three weeks after directing the company to halt access to its most advanced systems over national security concerns.
This decision comes as Washington continues to tighten scrutiny of advanced artificial intelligence technologies, reflecting growing concerns that powerful artificial intelligence models could be exploited by military or intelligence entities in countries such as China and Russia.
Export restrictions were lifted
Anthropic confirmed on Tuesday that the US Commerce Department has withdrawn export controls on the Fable and Mythos models.
Read also | Chinese AI company Z.ai targets Anthropic’s Cloud Mythos
The company said it will begin restoring access to models starting Wednesday.
“We will begin restoring access tomorrow,” Anthropic said in a statement on X.
The move follows a previous order issued on June 12 that forced the company to suspend access to the Mythos 5 and Fable 5 models due to export control requirements.
The government relaxes its previous order
The latest decision extends the partial relaxation that was announced last week.
The US government on Friday allowed Anthropic to make Mythos 5 available to a limited group of “trusted” US organizations, easing some previous restrictions while broader export controls remained in place.
Read also | US allows limited access to Anthropic’s Mythos AI model after blocking it over national security concerns
Now, according to a letter from US Commerce Secretary Howard Lutnick seen by Reuters, those export controls have been withdrawn completely.
Lutnick wrote that a license is no longer required to export Mythos or Fable models.
Security Commitments of Anthropy
The withdrawal of restrictions comes after Anthropic agreed to work closely with the US government on security measures for artificial intelligence.
“Anthropic has agreed to proactively disclose and address security risks associated with models; work diligently with the US government on protocols, standards, and releases for Mythos, Fable, and future models; and report any malicious activity to the US government,” Lutnick said.
The US government’s role in determining which organizations have access to advanced AI models has drawn criticism within the technology industry.
Sam Altman, CEO of OpenAI, questioned this approach last week in a post on X.
“Comprehensive safety tests are not a bad idea. I don’t like the idea of the government selecting customers.”
(With Reuters inputs)