The Mythos model is reportedly accessed by unauthorized users. Anthropic has claimed that its technology is extremely powerful.

Golden Finance reports that on April 22, according to an informed source and documents reviewed by the media, a small number of unauthorized users have gained access to Anthropic PBC’s new-generation Mythos AI model. Anthropic previously stated that the technology behind this model is extremely powerful and may even be used to launch dangerous cyberattacks. People who requested anonymity for fear of retaliation said that on the very same day Anthropic first announced plans to open testing of the Mythos model to a small number of companies, a small number of users on a private online forum successfully accessed Mythos. The informed source said that since then, this group has been regularly using Mythos, but has not used it for cybersecurity purposes. The informed source also provided the media with screenshots and live model demonstrations to substantiate the above claims. This unauthorized access had not been disclosed previously, highlighting the challenges Anthropic faces in preventing the diffusion of its most powerful—and possibly most dangerous—technology beyond authorized partners. The incident also raises further questions: whether anyone else is using Mythos without permission, and what their purposes might be. (Zhitong Finance)