Anthropic has launched its new cybersecurity AI model, Claude Mythos Preview, to a select group of major customers including Amazon, Apple, and Microsoft. This limited release follows recent data leaks that exposed details about the project and the company's internal source code.
The model is restricted to vetted organizations like Broadcom, Cisco, and CrowdStrike, with Anthropic also in discussions with the US government regarding its potential use. Anthropic cited "human error" as the cause for both the Mythos description leak and the subsequent exposure of its personal assistant's source code.
Mythos, designed as a "general purpose" AI, possesses the ability to identify cyber vulnerabilities at a scale beyond human capacity. However, Anthropic noted its potential to also develop methods to exploit these weaknesses, posing a risk if misused. The company stated the model could "reshape" cybersecurity practices and does not plan a broad release due to these potent capabilities.