Researchers from the Carnegie Mellon AI Safety Institute (CMU-AISI) have published a paper detailing a novel jailbreak technique called "Sleepwalker." This method uses a complex sequence of nested, se
Your 3-minute AI news briefing for Tuesday, July 7, 2026. The top stories in artificial intelligence, delivered daily.
ServiceNow has released EVA-Bench 2.0, a new benchmark featuring 121 tools and 213 scenarios to test advanced LLM agent capabilities. How will this change how we evaluate AI performance in complex, multi-step tasks?
Google is tackling hybrid work fatigue with a new 3D telepresence system for Google Beam, projecting life-size, photorealistic colleagues into meetings. Can this AI-driven tech finally make virtual collaboration feel real?
NVIDIA has launched Morpheus Guard, an open-source cybersecurity framework designed to provide real-time threat detection for AI-powered applications. Built on top of their Morpheus AI application fra
Researchers at Carnegie Mellon University's CyLab have published a paper detailing a novel jailbreak technique called 'Model Splicing'. The attack involves crafting prompts that force a large language
Cognition Labs, the company behind the popular AI software engineer 'Devin', announced a significant data breach that occurred in late May 2026. Attackers exploited a zero-day vulnerability in their p
The European Union's AI Act has officially entered its first enforcement phase, with the AI Office issuing new binding directives for companies deploying 'high-risk' AI systems. Effective immediately,
Researchers from Carnegie Mellon University have published a paper detailing a novel jailbreak technique named 'Sleepwalker'. This universal attack bypasses the safety alignment of major large languag
Your 3-minute AI news briefing for Monday, July 6, 2026. The top stories in artificial intelligence, delivered daily.
Google's I/O 2026 keynote revealed Gemini Omni, a powerful new AI model for real-time video understanding. Learn about Project Antigravity and major AI updates. What does this mean for developers?
A new framework proposes three 'inverse laws' of AI, exposing deep-seated issues with data bias, hallucination, and model disposability. Are today's AI systems fundamentally flawed?