Wednesday, June 10, 2026

Also

• Microsoft's open source tools were hacked to steal AI developer credentials — Security reminder: keep your keys rotated and monitor for suspicious activity.
• Judge cancels trial after learning both lawyers used AI—and didn't disclose it — A wild but important reminder that AI disclosure still matters in regulated spaces.
• Amazon employees mock the company's AI rollout on Slack—'Sloppenheimer' — Internal pushback on AI features that don't work as promised—tells you something about execution.
• AI misidentification results in wrongful arrest; victim seeks justice — High-stakes failure of facial recognition AI in law enforcement—real harm, real accountability.
• Where is the AI jobs crisis? (Apollo analysis) — Data-driven look at whether AI has actually displaced workers yet—spoiler: not at scale.
• Is Grep All You Need? How Agent Harnesses Reshape Agentic Search — Research on efficient retrieval for AI agents—practical if you're building search-heavy systems.
• Can LLMs Beat Classical Hyperparameter Optimization? — LLMs applied to ML tuning problems—interesting if you're experimenting with AutoML.
• Unified Controllable and Faithful Text-to-CAD Generation with LLMs — LLMs generating CAD models from natural language—useful for design automation workflows.
• Rich Sutton on AI creativity and discovery — Thoughtful thread from an AI pioneer on what separates capable systems from creative ones.