Late last week, Anthropic made the decision to take its latest AI models, Claude Fable 5 and Mythos 5, offline following a directive from the U.S. government. This directive prohibits any foreign nationals from utilizing these services, prompting the company to engage in discussions with the White House in hopes of reaching an agreement that would permit the reinstatement of their offerings.
Since its launch in April, Mythos has been touted for its impressive capabilities, particularly in identifying software vulnerabilities that can aid cybersecurity professionals in fortifying systems. However, the same technology poses risks, as it can also be exploited by malicious actors. Anthropic acknowledged this critical issue in a blog post, stating, "A great deal of advanced usage of AI models is dual use: the same queries that are beneficial in the hands of cybersecurity professionals and biology researchers could be dangerous if available to malicious actors."
Initially, Anthropic released a version of Mythos called Mythos Preview to a select group as part of Project Glasswing, a collaborative effort aimed at addressing these dual-use challenges. Last week, Mythos 5 was privately shared with this consortium, while Claude Fable 5—a model related to Mythos—was made available to the general public with limitations on its ability to respond to inquiries regarding biology and cybersecurity. This situation underscores the ongoing debate and concern regarding the responsible development and deployment of advanced AI technologies.
