New AI can actively exploit security gaps, Jörn Müller-Quade in interview with WDR 5
The AI company Anthropic has announced a new AI language model that could present a potential threat to public cybersecurity. According to the developers, their new model, “Claude Mythos Preview,” not only identified security vulnerabilities in commonly used software but also discovered how to actively exploit them. Due to these risks, Anthropic will not release the AI for the time being.
Prof. Dr. Jörn Müller-Quade, Professor of Cryptography and Security at the Karlsruhe Institute of Technology (KIT) and spokesperson of KASTEL Security Research Labs, commented on this in an interview with WDR 5.