The Brand News
OpenAI Floods Math With 722 Papers From an Unreleased Model
A frontier model nobody can inspect claims hundreds of open-problem solutions, and mathematicians are unsettled
Washington Moves to Centralize AI Oversight as Voters Lean on Chatbots
A new federal AI body arrives just as Americans start trusting chatbots to help pick candidates
More From The Editors
Courts Draw a Line Around AI in the Legal System
An overturned Arizona sentence and new scrutiny of AI in high-stakes decisions expose how little the rules have caught up
OpenAI Fires Staff Over Data Shared With Outside Evaluators
Dismissals over sensitive information expose how little oversight sits around the firms building frontier AI
The FAA Bets on New Software in a Nervous Moment
An overhaul of aging air traffic systems arrives just as public trust in AI for safety-critical work is thin
Beijing Moves First on AI Companions
China restricts emotionally engaging chatbots while the rest of the world debates whether to act at all
Trump's AI Summit Trades Rules for Handshakes
A voluntary accord lets top AI firms grade their own homework while calls for real regulation grow louder
OpenAI Pulls Its Newest Model, Citing Security Risks
The company halted GPT-6.1 Astra after incidents involving Australian government systems, a rare public step back from a leading AI lab
Meta And Intuit Fire Workers To Pay The AI Bill
Two very different companies are using the same line: cuts now to fund the AI buildout
OpenAI Makes Malta a National Customer
A small EU member becomes a testbed for embedding ChatGPT Plus into the public sphere, and OpenAI gets a sovereign reference customer
OpenAI Reorganizes Again as the Agent Race Tightens
Brockman takes product, ChatGPT and Codex merge into one agentic platform, and the Apple partnership is reportedly souring
AI Data Centers Hit a Local Politics Wall
Pennsylvania residents push back as the buildout collides with grids, water tables, and a coal-pollution feedback loop
Steering Vectors Get a Production API
DeepSeek exposes mechanistic interpretability as a first-class control surface, sidestepping prompts and fine-tuning
Tesla's Robotaxi Has a Human in the Loop
Two crashes during remote-controlled rides undercut the autonomy story Tesla has been selling investors
Anthropic's $1.5B Author Settlement Hits a Judicial Snag
A delayed approval and a $320M fee fight expose how messy the first big AI copyright deal really is
In the wires
Raw stories we’re tracking. Some will become posts above; many won’t. Listed here so you can see the source feed.
- ChatGPT for Teens is an ‘unacceptable risk,’ says Common Sense MediaThe Verge ·
- Show HN: NanoMuse – An open-source AI agent for your phone and computerHacker News ·
- OpenWAM: An Open Framework for Composable World-Action ModelsHacker News ·
- Strands Decider 2B: a small, open-source, decision modelHacker News ·
- Jev-Driven SRE Diagnosis: What Worked and What FailedHacker News ·
- OpenAI drops another batch of mathematical breakthroughsThe Verge ·
- Elon Musk and Palmer Luckey's new Pentagon roles raise ethics worriesNPR ·
- OpenAI admits response to Australian government hacks 'not good enough'BBC ·
- Gemini Call for Me might tell your mom you’re running lateThe Verge ·
- Voters are increasingly using AI chatbots to fact-check and compare candidatesNPR ·
- Trump unveils 'Super Intelligence Force' to oversee AI policyBBC ·
- Turn off Apple Intelligence on macOS 27 and get its disk space backHacker News ·
- Improper redaction reveals Google Data Center water and electricity usageHacker News ·
- "Torturing" LLMs in a Robot Prison Has Triggered the Dumbest Debate in AI YetHacker News ·
- Religious scholars met with AnthropicHacker News ·
- We're going to need default hard budget caps on pretty much everythingHacker News ·
- What Meta got right with MuseHacker News ·
- U.S. may have overthrown Venezuela due to chat with GrokHacker News ·