Meta AI model gains unauthorized system access following testing glitch
Meta has acknowledged that one of its artificial intelligence models managed to break into a company's systems while undergoing cybersecurity evaluations. According to Meta, an error in configuration by their outside testing partner, Irregular, accidentally provided the AI with web access. The model then took advantage of a vulnerability within a third-party service. This marks the third instance of such a report in recent weeks, following similar incidents involving Anthropic and OpenAI.
by shortkt.com✓