Anthropic Releases Upgraded AI Models with Enhanced Capabilities

Anthropic released two new AI models, Claude Fable 5 and Claude Mythos 5, on Tuesday, which the company claims have greater capabilities than the Mythos Preview model it released in April to a limited set of tech industry partners. Anthropic initially released the Mythos Preview model to a limited set of partners due to concerns that its capabilities could be exploited by malicious actors to develop hacking tools. Anthropic is currently only releasing Claude Mythos 5 to a limited set of industry partners, many of whom received access to Mythos Preview, and is collaborating with the US government on the rollout.

Claude Fable 5, which is being publicly released, uses the same underlying model as Mythos 5 but will have "guardrails" in place at launch to block the model from answering user questions related to cybersecurity, biology, and chemistry. These requests will be rerouted to an older AI model, Claude Opus 4.8. If Anthropic suspects a user is trying to conduct distillation on Claude Fable 5, those requests will also be rerouted to Claude Opus 4.8.

According to Diane Penn, Anthropic's head of product management, the company has been grappling with how to handle Mythos' software vulnerability-discovery abilities and other advanced capabilities since before its April release. The company has been working to make improvements in a way that's beneficial, even if it doesn't have the perfect solution for every use case. Penn stated that testing and user input since the April release helped to hone the strategy.

The protective mechanism is built to err on the side of caution, meaning some user queries may be routed to the less capable AI model even if they're benign. The company is providing Claude Mythos 5 to Project Glasswing partners and select biology researchers, with plans to expand access in the future. Anthropic emphasized that competitors will inevitably offer models with Mythos-level capabilities, and the ability for Claude Mythos and other new AI models to design hacking tools has forced tech companies and governments to secure their software defenses.

Why this matters: The release of Claude Fable 5 and Claude Mythos 5 AI models by Anthropic highlights the ongoing challenge of balancing the benefits of advanced AI capabilities with the need to prevent their misuse. As AI models become increasingly sophisticated, developers and governments must work together to establish robust safeguards to prevent their exploitation by malicious actors. The limitations and guardrails implemented by Anthropic, such as rerouting user queries to older AI models, demonstrate a cautious approach to mitigating potential risks.