Who's Watching the Watchers

A person ran their MRI results through Claude and found it surfaced details their radiologist had mentioned but not fully explained. This is not a replacement for a doctor, but it is a preview of what it looks like when AI becomes the second reader you can actually afford.

A Brown University professor flagged what appears to be widespread AI-written submissions on a single exam, enough to make it a public story. If you work in education, or manage people whose work you review, this is the moment where 'assume good faith' gets complicated.

A security-focused team at Semgrep ran their own benchmarks and found GLM 5.2, a Chinese open model, outperforming Claude on cybersecurity tasks. Benchmarks are always partial pictures, but this one got 875 upvotes because it came from a team with a specific, practical use case rather than a lab trying to sell something.

An open GitHub issue shows that OpenAI's Codex still has no reliable way to tell it which files to keep its hands off. If you are using an AI coding tool on a project with secrets, credentials, or private data sitting in the same folder, this is worth knowing before it becomes a problem.

Jon Udell flipped the phrase 'human in the loop' to 'agent in the loop,' arguing the framing matters because it changes who we think of as being in charge. Small language shift, but worth sitting with if you are building workflows where AI takes actions on your behalf.