Anthropic keeps better model 2 internal over safety concerns
Anthropic has built a next-generation AI called Model 2 that is even better than its current Mythos model, with noticeable improvement on various internal tasks and is being used for coding, agentic work and data generation.
But here is the catch: you will not get to try it any time soon. The company is keeping Model 2 under wraps because it is worried about safety risks if it goes public.
Other AI firms pause model releases
Even though Model 2 did well in tests, Anthropic says the jump is not as big as its last upgrade, and recent cybersecurity issues have made it more cautious.
Its risk rating for high-stakes situations went from "very low" to just "low."
So, Model 2 will stay internal for now, while Mythos 5 is being used heavily inside Anthropic.
Anthropic is not alone: other AI companies like OpenAI are also hitting pause on releasing their most powerful models until safety catches up.