White House Considers Pre-Release Reviews for Anthropic Mythos

The White House is reconsidering how to oversee advanced AI models due to concerns about Anthropic’s new Mythos model, which could enable cyberattacks. This comes as discussions begin on whether to implement pre-release reviews for high-risk AI systems. The Trump administration is weighing options like formal vetting processes and government testing for these technologies. This marks a shift from the previous stance, where oversight was largely removed when Trump took office in 2025. At that time, he revoked key regulations aimed at ensuring safe AI development. What's interesting is that Anthropic claims Mythos can identify thousands of severe software vulnerabilities, outpacing even skilled human coders. They have not released the model publicly but have launched Project Glasswing, investing $100 million in credits for select tech companies to use Mythos defensively. This matters because as AI capabilities grow, so do the risks, making it crucial for the government to find a balance between innovation and safety in technology that impacts us all.