Anthropic’s yet-to-be-released AI model, “Claude Mythos,” was inadvertently revealed due to a data breach. The company admitted that the leak stemmed from the model’s descriptions being stored in a publicly accessible data repository, and it hailed the model as a “transformative breakthrough” and the “most advanced model ever created.” However, the leaked documents also highlighted that the model could introduce “unprecedented cybersecurity threats.”

As a result, cybersecurity stocks experienced a sharp decline across the board in pre-market trading on Friday, with the downward trend accelerating after the market opened. The slump in cybersecurity stocks underscores broader market anxieties about AI disrupting the traditional security software sector, as investors fear that powerful AI models may diminish demand for conventional cybersecurity solutions among corporate clients.

The contrasting interpretations of the leaked documents, compounded by thin pre-market liquidity, further intensified selling pressure in the sector. At present, the official launch date and technical specifications of Claude Mythos remain undisclosed, and the market will closely track further developments and official announcements.