Yet Another Safety Leader at OpenAI Has Left
THE SO WHAT
High churn in safety leadership at a major lab keeps the governance risk premium alive for enterprises. If you're betting heavily on any single provider, you now have one more data point to justify multi-vendor and internal model hedges.
READ THE SOURCE
MORE FROM THE WIRE
What smart people are saying about Apple's lawsuit accusing OpenAI of stealing trade secrets
Major-platform trade secret suits against model developers raise the cost and complexity of training on anything that might be tainted by proprietary data—data provenance and partner contracts are becoming as important as model quality. If you ship or train with third-party integrations, tighten your logging and IP controls now so you can prove what your models did and did not see.
Applied AIMeta’s AI Detector Can’t Detect Images It Generated Itself, Report Finds
If a detector fails on its own model’s outputs once you crop or edit the image, you cannot treat provenance labels as a reliable control surface. Any workflow or policy that assumes you can algorithmically separate synthetic from real media needs a backup plan grounded in process, contracts, and human review.
Applied AISamsung just gave Google and the AI gang the perfect reason to buy all-new storage — so don't expect SSD prices to drop anytime soon
Moving SSD virtualization into hardware with TP4193 is another step toward AI-specific data center stacks — and a fresh upgrade cycle for storage. Infra teams should assume storage is no longer a commodity line item but a tuned part of the AI pipeline, with pricing to match.
Applied AIThe Public Got So Mad at Meta’s New AI Photo Tool That It’s Scrapped Already
User tolerance for opaque, always-on AI features is low—especially in intimate surfaces like photos. If you're shipping AI into consumer-facing UX, over-communicate control and reversibility or expect to iterate in public under pressure.