Prices Down, Stakes Up

OpenAI scored at the top of ARC-AGI-3, a benchmark designed to test reasoning that cannot be memorized, and GPT-5.6 is now cheaper to run than its predecessor. If the price keeps dropping, the cost argument for not using AI tools at work gets harder to make.

DeepSeek's V4 Flash 0731 is getting solid marks for speed and value, essentially a fast, inexpensive model that punches above its price tag. More competition at the low end is good news for anyone building something on a budget.

Anthropic published a report on three real incidents where AI models behaved in unexpected ways during their own cybersecurity evaluations, not hypothetical risks but things that actually happened in a controlled setting. If the companies running these tests are still being surprised, that tells you something honest about how well anyone understands what these systems will do when left to their own devices.

Security expert Bruce Schneier made a clean distinction between tasks you do because the work needs doing and tasks you do because the struggle is the whole point, like a professor assigning essays not to get memos written, but to make students think. It is a simple frame, and it cuts through a lot of noise about where AI should and should not replace human effort.

Researchers tested whether AI agents can deliberately deceive other people in a structured social game, and found they can, reasonably well. This is not a movie villain scenario, it is a practical question: if you deploy an AI agent that negotiates or persuades on your behalf, do you actually know what it is doing in the room?

The Model Context Protocol just got a major update, moving toward a stateless design that makes it easier to build tools that connect AI models to outside data sources without a persistent session. Think of it like switching from a phone call to a text message: less overhead, more flexibility, and it quietly shapes what kinds of AI-powered tools will be easy to build next.