Artificial intelligence company Anthropic launched its new Fable5 model from the Claude Mythos series to the public on Tuesday. This is the first public appearance of this cutting-edge model family, which had previously been strictly limited to internal testing among a few partners due to cybersecurity risks.
To balance technological openness with security measures, the release of the new model adopts a tiered filtering mechanism. For queries related to highly sensitive areas such as cybersecurity, biology, or chemistry, the system automatically redirects them to lower-tier models with reduced capabilities.
Project Glasswing Initiative Collaborates with Government and Enterprises
Alongside the release of the general version, Anthropic provided unrestricted high-tier versions to signed companies and cybersecurity partners involved in the Project Glasswing initiative. Currently, this special global collaboration network has expanded to cover approximately 200 organizations across more than 15 countries.
At the same time, the U.S. government has also been deeply involved in the security evaluation of the model. Given the strong capability of the Mythos model to identify critical infrastructure vulnerabilities, the White House has established a routine mechanism to conduct security reviews of top AI models before their public release.
High Costs and Intense Industry Competition
As one of the most advanced AI technologies, Fable5's pricing has doubled compared to previous versions, reflecting the significant financial pressure the company faces. To obtain sufficient computing power, Anthropic currently pays substantial data center rental fees to Elon Musk's xAI every month.
In the context of increasingly fierce market competition, both Anthropic and its main competitor OpenAI have recently submitted plans for initial public offerings (IPOs). This capital boom driven by artificial intelligence is pushing the business competition in the tech industry to an entirely new historical level.
