2026-09-10 · Fuente：SiliconANGLE p2
Google Threat Intelligence Group says a financially motivated actor used a multi-agent AI framework to compromise thousands of credentials in under six hours, in a campaign traced by Mandiant.
Google LLC's Google Threat Intelligence Group disclosed on September 8, 2026 that threat actors deployed a multi-agent artificial intelligence framework to compromise thousands of credentials in under six hours. The findings appear in a report released today by the company's threat intelligence unit, which describes an operation in which autonomous AI agents carried out credential theft at a pace that compresses the timeline of a conventional intrusion campaign.
Mandiant investigators traced the campaign to a suspected financially motivated actor, according to the report. The intrusion began when the attacker broke into an organization's cloud infrastructure. From that initial foothold, the actor assembled an autonomous framework, which then proceeded to compromise thousands of credentials, completing the entire sequence in less than six hours.
For defenders, the compressed timeline is the central problem. An operation that moves from an initial cloud breach to mass credential compromise in under six hours leaves little room for the detection-and-response cycles around which enterprise security operations are usually built. Autonomous agents do not pause between steps for a human operator to weigh options, which means the window to interrupt an attack in progress shrinks accordingly.
The case also illustrates how the economics of intrusion work shift when AI agents absorb the operational load. Mandiant's assessment characterized the actor as financially motivated, yet the multi-agent framework allowed credential theft to proceed at industrial pace without continuous hands-on control. For organizations running cloud infrastructure, the disclosure is a reminder that a single entry point can cascade into thousands of stolen credentials within hours, strengthening the argument for automated detection on cloud platforms and closer oversight of agentic AI tooling inside enterprise environments.
Etiquetas: Google · Google Threat Intelligence Group · Mandiant · AI Agents · Cybersecurity · Credential Theft
萬安算交所 AIXX · AIXX · Exchange verificado · Ir al mercado