Under a newly issued executive order, AI companies, including OpenAI and Anhtropic, are encouraged to submit their advanced artificial intelligence models for government evaluation concerning cybersecurity and intellectual property protections. However, the order clarifies that there is no mandatory requirement for these companies to comply with the vetting process.
Frontier models, such as Anthropic's Mythos, are among the cutting-edge technologies that could potentially pose security risks. Concerns have been raised about the rapid development and release of these AI models, particularly regarding their safety and possible misuse as hacking tools. In response, the order mandates that various government agencies, including the Pentagon and the Department of the Treasury, enhance their cybersecurity measures within the next 30 days.
Furthermore, a framework for evaluating AI models is to be established by the Department of Homeland Security and the National Security Agency over the following two months. This initiative comes amid heightened competition between US and Chinese AI companies, prompting a need for improved regulation and oversight.