An AI-assisted newsroom with a human journalist always in the loop.
Lead story
Two days after Anthropic's Claude Fable 5.1, OpenAI says its new model beats it on the scientific-agent benchmark at the same $10 and $50 list price. Its own system card says the model's reasoning got harder to monitor.
Edited by Trevor Brown · Sep 4, 2026 · frontier labs
OpenAI released GPT-6 Astra on September 3, two days after Anthropic launched Claude Fable 5.1. Its launch page compares the two by name.
Edited by Trevor Brownfrontier labsSep 1, 2026
Astra is the first model OpenAI has designated Critical for cyber capability under its Preparedness Framework. It is releasing it behind heavier friction, and warning users the safeguards can sometimes stop legitimate work.
Edited by Trevor Brownfrontier labsAug 26, 2026
Private Safety Processing, previewed August 19, is OpenAI's bid to spot misuse across related interactions while keeping customer content away from its staff — squaring expanding safety monitoring with a zero-retention promise.
frontier labsSep 8, 2026
The company says its AI agents produced a proof about when fluid equations break down, a potential milestone for AI in mathematics that still needs outside scrutiny.
frontier labsSep 1, 2026
Anthropic says the new model more than doubles its predecessor on a scientific-agent benchmark and costs around 25% less to run. Its cyber filters also block 60% fewer benign requests, and now let the model hunt software vulnerabilities.
safetyAug 18, 2026
OpenAI says it already went two weeks without RL training on its deployment-bound models, and didn't say exactly when. Its largest planned frontier run is still frozen while it hardens security, expands monitoring and gathers evidence of alignment.
frontier labsAug 13, 2026
SpaceXAI shipped a new flagship four weeks after the last one, left headline pricing at $2 in, $6 out and raised the cached-input rate by two-thirds. Its own comparison table shows a model that matches rivals on a composite index, leads on some benchmarks and trails on others.
The context engine
Beyond the daily file — a reference desk for the AI industry. Check the profiles, follow the timeline, and see the sources behind every claim.
Analyst · Wharton professor focused on AI use and work.
Existential AI risk is obviously critical, but it is not the only AI thing that requires policy. I worry it will become the sole focus of AI discussions. We don’t need better models for AI to have wide impacts on jobs & society and we need to be preparing to encourage good outcomes & mitigate bad.
Analyst · Interconnects writer; former Ai2/OLMo and Hugging Face.
Open-Source AI & Open Models Reading List How to get up to speed on open models and their implications. www.interconnects.ai/p/open-sourc...
Anthropic says it will watermark Claude's output. Nothing is marked yet.
OpenAI ships a cyber model trained to refuse less
OpenAI can't rule out "critical" cyber capability in Astra, and is slowing the model down
Alibaba says its most capable model ever goes open-weights next week
Anthropic answers the open-weights letter: 'never advocated for a ban'
What we learned mapping a year's worth of AI-enabled cyber threats
MAI-Code-1-Flash is now available for GitHub Copilot
