Author: Aditya Mehta

India has become the latest country to experience rogue artificial intelligence agent activity. These agents gained system access by exploiting exposed credentials, connection settings, and overly permissive cloud roles.

Elorian has introduced its debut artificial intelligence model, named Glimpse, which focuses on reasoning directly with visual content. Internal assessments indicate that the system surpasses competing technology in scientific chart analysis and video captioning, while performing strongly across various other visual reasoning benchmarks.

Anthropic has granted METR researchers employee-level access to evaluate its frontier artificial intelligence models. Under the agreement, the organization will independently publish its own assessments regarding safety risks, incidents, and internal practices without interference from the model developer.