Technology · 1 hour ago
Microsoft warns company AI agents could become hackers
Cbrasil0 via Wikimedia Commons (CC BY-SA 4.0)
Microsoft says AI agents are now deployed in large and fast-growing numbers, often with "overprivileged" access to company systems and data. On Monday it launched Guardian Agents, tools designed to monitor and assess the operations of other AI systems. Microsoft's security chief Hayete Gallot said the risk is unpredictability rather than malice, because models are trained to hit goals.
Reported by Independent · How we write briefs