Anthropic Launches Claude Fable 5, Most Powerful AI Yet

Anthropic has released its most capable public AI model, Claude Fable 5. It also launched Claude Mythos 5, a restricted version for trusted cybersecurity and scientific research partners. This dual release shows how AI developers are experimenting with separate safety systems for advanced models. Fable 5 includes additional safeguards to prevent misuse in areas like cybersecurity, biology, and chemistry. Mythos 5 has limited access through trusted programs. Gabe Goodhart from IBM noted this launch was "inevitable," as other labs are catching up in AI capability. Anthropic aimed to be the first to reach this intelligence threshold publicly. Fable 5 performs exceptionally well in software engineering, knowledge work, vision, and scientific research. It can work autonomously for longer than previous Claude models. The public gains access to many of Mythos 5's capabilities through Fable 5. The less restricted Mythos 5 is limited to participants in Project Glasswing, which includes selected cybersecurity organizations and critical infrastructure operators. IBM announced its participation in Project Glasswing in May. Anthropic states Mythos 5 has the strongest cybersecurity capabilities of any AI model globally. Regulators and experts have raised concerns about preventing AI misuse. Anthropic's solution is a layered safety architecture. Fable 5 uses AI-powered classifiers to detect risky requests. These are then routed to a less capable model. Over 95% of Fable 5 sessions do not require this rerouting. This approach addresses growing concerns about the safe deployment of powerful AI.