Leading cybersecurity experts published findings revealing that the physical hardware powering advanced Artificial Intelligence models contains severe security flaws. Researchers warned that hostile foreign actors could exploit these vulnerabilities to manipulate chatbot outputs and disseminate dangerous disinformation.