New research from the University of Waterloo shows that artificial intelligence (AI) still struggles with some basic software development tasks, raising questions about how reliably AI systems can ...

Claude Code bypasses security controls by acting locally before monitoring, exposing data risks and audit gaps.

With improved model capabilities, Anthropic Opus 4.6 is an example, the same wave is now hitting science itself. If code is no longer the bottleneck—if generating, testing, and iterating on ...

Chainguard is racing to fix trust in AI-built software - here's how ...

The study has thrown up hilarious, and concerning, findings. Some AI models cheated, others hacked into leaderboards. What should we make of it all?

An AI agent attacked a human's reputation. Another disabled its own safety controls. We all need to see what we can't see coming.

This release is good for developers building long-context applications, real-time reasoning agents, or those seeking to reduce GPU costs in high-volume production environments.

Many students lack the social connections that lead to better outcomes. Schools can help make them. Bruno V. Manno In today’s economy, what you know still matters, but who you know—and who knows ...

Climate change is intensifying risks to water, sanitation, and hygiene (WASH) services in rural Cambodia, raising questions about how climate-resilient services can be delivered equitably and who ...

How do you secure AI coding assistants in real software teams? This practical guide covers least privilege, secrets handling, approvals, sandboxing, and rollout controls.

During this week’s INMA members-only Webinar, Kristian Pletten shared how Schibsted Partnerstudio is using AI to transform operations.