Anthropic's repeated emphasis on AI risks may have contributed to the US government's decision to ban foreign access to its latest models, Mythos and Fable. According to Financial Times analysis, 5 out of every 1,000 words used by Anthropic in 2026 related to risk, regulation, or restrictions, compared to 0.6 words per 1,000 for OpenAI. The export ban, which took effect last week, has drawn criticism from some technologists, including Yann LeCun, who accused Anthropic's CEO, Dario Amodei, of 'ridiculous fear-mongering' about AI's societal risks. The decision has raised concerns among European and Silicon Valley officials, who fear the Trump administration may expand restrictions on non-US access to frontier AI models. It also marks an early test of how the US intends to regulate increasingly powerful AI systems. Source: arstechnica

The Financial Times analysis compared the frequency of terms like 'risk,' 'safeguard,' and 'vulnerability' in statements from Anthropic and OpenAI. In 2026, 'risk' appeared 336 times in Anthropic's communications, while 'safeguard' and 'vulnerability' were used 121 and 128 times, respectively. In contrast, OpenAI used these terms 30, 33, and 10 times. Anthropic's CEO, Dario Amodei, argued in a blog post that regulators were moving too slowly on AI, citing 'very real risks to cyber security, creating the potential for disruption of the financial sector, critical infrastructure, and national security.' However, the FT noted that this post was less alarmist than some of Anthropic's earlier publications, which contained more negative language about AI's potential harms. Source: arstechnica

Anthropic has long positioned itself as the AI industry's conscience, frequently releasing research papers and calling for greater government oversight. The company initially limited access to Mythos to certain US organizations due to safety concerns and had been working with government officials on a controlled rollout before releasing it more widely earlier this month. Media coverage of Mythos, which was announced in April, was significantly higher than other models released this year, according to data from AlphaSense. Some industry figures, including David Sacks, have criticized Anthropic's handling of government discussions over the new models, suggesting the company downplayed concerns and forced the government to impose the ban. Source: arstechnica