By Emilio Crespo
The Trump administration has adopted a new and first standard to control the risks of Artificial Intelligence: national exclusivity. Its decision that only US citizens and no foreigners can use Anthropic’s most advanced – and dangerous – AI prevents even the multinational company’s own employees without a US passport from accessing or working on it.
This regulation was the first from the White House after announcing in May that it would supervise AI advances because they threaten national security.
The affected model is Fable 5, a light version of another superior one, Mythos, launched in April and which has been capable of hacking thousands of programs and devices considered safe. Anthropic limited access to Mythos to less than a hundred entities and its executives have warned of the danger it would have in the wrong hands. Washington suspects that a group linked to China wanted to use it.
Access to Fable, the limited version of Mythos, has been available to everyone for just two days. Furthermore, the national exclusivity that limits its access extends to Mythos, among whose privileged users were non-US entities.
Anthropic has criticized the measure of the US Government, with which it is confronted by canceled contracts and military use of its programs not authorized by the company. National exclusivity could be justified by many criteria, but hardly by preventing cybercriminals from doing their thing. But it is a new punishment for the company that has surprised the world with the power of its AI and that dared to confront Washington in defense of the rights of its citizens and those of other countries.