The White House imposes export restrictions on Anthropic’s Mythos model over concerns about access from China
A new Semafor report says the White House partly decided to place export restrictions on Anthropic’s...
AI BriefingSemafor reported that the White House decided to impose export restrictions on Anthropic’s Mythos model over concerns that China-linked groups could access it. Another risk is that external groups could steal the model’s capabilities through knowledge distillation. The U.S. Department of Commerce had previously directed Anthropic to disable Fable 5 and Mythos 5 after finding that jailbreaks could make the models disclose cybersecurity assistance. Anthropic countered that jailbreaks are not universal and that other publicly available models can provide similar capabilities. The restrictions will remain in place until the U.S. government strengthens its national security systems, which is expected to take place over the next few weeks. Anthropic acknowledged that no model provider can currently achieve perfect jailbreak prevention.
Reported by 4 other sources
- Who decides how dangerous AI is? US government restrictions on Anthropic Fable 5 spark controversyThe Verge: AI · June 18
- Explaining the political power struggle between the White House and Anthropic over the Fable modelsThe Verge: AI · June 18
- Anthropic’s enterprise AI subscription share surpassed OpenAI’s for the first time in May; Trump administration ban instead drives adoption to a record highTechCrunch: AI · June 17
- U.S. denies G7 allies special access to Anthropic’s Mythos 5 and Fable 5Rohan Paul · June 17
Source: Rohan Paul · x.com