The archive
Guides and explainers on artificial intelligence, newest first. Pick a topic to narrow things down.
OpenAI says it will watermark ChatGPT and Codex text output in the EU under the AI Act, while admitting detection drops to 17% after a quarter of words are swapped for synonyms.
6 min read
Anthropic says its new Claude Frontier Academy will train 10,000 "Frontier Deployed Engineers" by the end of 2027, with first cohorts from Accenture, McKinsey, Morgan Stanley and Novo Nordisk.
5 min read
An MIT profile of associate professor Cathy Wu traces years of failed RL experiments to a 2023 result: train on the 10% of problems that generalize, and skip the rest. Up to 30x more efficient.
5 min read
OpenAI's new GPT-6 developer guide lays out three tiers—Astra, Sol and Luna—with a 100x price spread, plus steering, async tools and delegation for tasks that run for hours.
6 min read
MIT's Martin Trust Center has launched Dear Dreamer, a free AI-powered platform that adapts its 24-step Disciplined Entrepreneurship framework for middle and high school students.
4 min read
A CSAIL-led tool pairs Microsoft's TRELLIS 3D generator with GPT-4 so novices can spot and fix structural flaws before printing. Nearly 80% of generated test models were flawed.
5 min read
A new essay published on OpenAI's Intelligence Age platform argues that frontier AI's biggest contribution may be executing ideas, not generating them — and that execution, not genius, is the binding constraint.
5 min read
OpenAI says Albertsons is expanding its partnership to bring a Safeway shopping experience into ChatGPT and roll out ChatGPT Enterprise internally. Metrics were not disclosed.
5 min read
Barclays says it will extend Anthropic's Claude Code to 50% of its developers by the end of 2026 and most engineers in 2027, alongside live deployments in UK retail support and Global Markets.
5 min read
Harvard's Matthew Schwartz says he stopped asking Claude to act like a scientist and built a harness around what it does well. The results span 18 fields — and reveal where agents still fail.
6 min read
OpenAI says a Denver family social club cut grant applications from three days to two hours using ChatGPT Work. The case study is a useful look at small-business AI adoption, and at what these metrics leave out.
5 min read
Google DeepMind's new frontier model claims a 1M-token output limit and state-of-the-art scores on software engineering and security benchmarks, but access starts with a small group of vetted defenders.
5 min read