Microsoft: “Frontier AI models, regardless of their origin, cannot be trusted”

Microsoft says frontier AI models cannot be trusted and urges companies to treat powerful AI agents as potential insider threats.

Preview courtesy of Neowin. The full article opens on their site.