Anthropic has released two new AI models: Claude Fable 5 and Claude Mythos 5. While the former is publicly available with safety measures in place, the latter remains exclusive to a select group of tech partners.

The introduction of these models comes amid concerns about cybersecurity vulnerabilities that advanced AI could exploit. Anthropic has implemented guardrails on Fable 5 to prevent it from generating certain types of content related to hacking and scientific research, rerouting such queries to an older model named Claude Opus 4.8.

In an interview with WIRED, Diane Penn, head of product management at Anthropic, explained that the company is working to find a balance between innovation and security. 'We're trying to make improvements in a way that's beneficial, even if we don't have the perfect solution for every use case to start,' she said.

While these measures offer some protection, they also highlight the ongoing tension between AI development and cybersecurity challenges. Anthropic continues to emphasize its commitment to expanding access but acknowledges the need for robust safeguards in place.