Anthropic's 'Model 2': More Powerful Than Mythos, Unreleased

Anthropic is developing an advanced AI model, internally called 'Model 2,' which is more capable than its current top-end Mythos model. However, the company has no plans to release it publicly. What's interesting is that Model 2 shows noticeable improvement on various internal tasks. It's being used for coding, agentic work, and data generation within Anthropic. The company states Model 2 is an experimental system, part of its standard research and development process. Anthropic also recently changed its risk assessment for AI misalignment, moving it from "very low" to "low" after cybersecurity incidents. This suggests a growing caution within the industry regarding advanced AI. This matters because it highlights the balance between AI development and the careful consideration of its potential risks.