🤖 OpenAI Math Claim Draws Scrutiny

The reported result involved an internal model more capable than GPT-6 Astra, solving the Navier-Stokes problem with 10,000 parallel agents. In September, OpenAI said it would not claim the result. The problem is one of 7 Millennium Prize Problems, carrying a $1M prize.

1 day later, NYU professor Tristan Buckmaster and Anthropic researcher Levent Alpoge published work on a related problem using Codex and Claude. Buckmaster says OpenAI followed the same route.

The dispute concerns training data and quiet edits to published papers. OpenAI denies using specific user data but says anonymized data may have improved its models.

📊@tech

🤖 Anthropic releases cheaper Haiku 5.5

Anthropic’s first Haiku update in a year costs up to 20 times less than Sonnet 5.5. It is already available to all users.

The model handles routine work, including summaries and document tasks, plus search support for Opus and Sonnet. Benchmarks put it ahead of GPT-6 Luna.

📊@tech

🖤 Ben Affleck Writes Python

The actor writes Python scripts, understands convolutional neural networks, and works extensively with GPUs.

He learned machine learning through visual effects, then founded InterPositive, an AI startup later acquired by Netflix. He says he can call Google or OpenAI as Ben Affleck and sometimes get a look at their latest models.

📊@tech

💠 Tesla Expands Car-to-Home Backup

Owners of Model 3 and Model Y in Premium and Performance versions can now use the car to power their home during an outage. Tesla says the backup can last more than 2 days. Cybertruck can run a home for more than 3 days.

The feature requires Powerwall 3. Tesla guarantees support only for cars ordered from October 1, since older vehicles may lack the required PCS2 Lite inverter. It is currently limited to the US and Puerto Rico.

Model S and Model X are excluded, while grid support remains a Cybertruck feature. Kia and GM already offer home backup from their vehicles, and Nissan Leaf has fed energy into the grid for years.

📊@tech

🇺🇸 US Blocks Big Tech Green Cards

The US Labor Department will stop reviewing all current and new applications. H-1B employees at affected companies will no longer receive work-based US residence.

The measure covers Microsoft, Adobe, and other big tech firms. The White House wants to target foreign students at Harvard, Yale, and Stanford next.

📊@tech

☀️ Blue Origin’s Lunar Power Tower

Blue Origin’s 20-meter Power Tower is designed to deploy after landing. The Mark 1 lander extends a spiral mast, then unrolls a folded solar panel from its top, like window blinds. The panel tracks the Sun on its own.

At the lunar south pole, the Sun stays near the horizon and shines from the side. A flat panel on the lander would see little of that light, while the vertical design is built for it.

One tower should produce more than 10 kW. Solar power could be available for 44% to 100% of the time, depending on the season. Several towers could form a microgrid for lunar vehicles and base systems.

📊@tech

🖥 Microsoft Rebuilds Windows for AI Agents

Its Hybrid Intelligence system will route each task between on-device models and the cloud. On powerful PCs, models with hundreds of billions of parameters can run locally. Microsoft showed MAI Code 1.1 Flash with 137 billion parameters and a 256K context.

The target is autonomous agents that can operate on the computer and write code, with access to files and apps.

Execution Containers isolate the agent and let users grant separate access to selected files or the internet. They already support several tools, while Claude Code is planned for later.

📊@tech

🤖 AHM Calls for OpenAI Boycott

The Association for Humanistic Mathematics called for mathematicians to stop working with OpenAI after the company released 722 AI-generated mathematical manuscripts covering 372 groups of results, including claimed solutions to longstanding problems.

Terence Tao published the statement as a guest post on his blog. AHM says OpenAI ignored warnings against using unsolved problems to test internal AI models, and called the release a show of strength rather than scientific work.

OpenAI says the aim is to advance mathematical research and let researchers examine the results. Researchers still have to verify the papers and assess their value.

📊@tech

🤖 Anthropic Adds a Cruelty Rule for Claude

From November 12, 2026, Anthropic’s policy will ban “systematic and unjustified abusive or cruel behavior” toward its models. It applies only in extreme cases involving repeated cruelty without an apparent purpose.

Ordinary frustration and resistance are excluded. So are dark creative texts and model testing. Claude can already end chats with repeatedly abusive users on and Claude Code. Anthropic says this will remain the main enforcement tool.

The update also extends the weapons ban to software used by weapons and drone armament. It clarifies that tracking people without consent is prohibited and adds safety requirements when Claude controls equipment that could harm people.

📊@tech

🚀 SpaceX Targets the US Mobile Market

SpaceX wants Starlink Mobile to become a major US mobile operator. It bought licenses for up to 14 MHz of paired spectrum in the 800 MHz band, where signals can pass through walls and buildings.

The planned service combines satellites that connect directly to smartphones with a conventional terrestrial network. SpaceX says it would be the first operator to unite satellite and terrestrial spectrum in one network.

The FCC has approved 15,000 second-generation Starlink Mobile satellites. SpaceX says their capacity will exceed today’s level by more than 100 times. The spectrum deal still needs FCC approval, so deployment starts only after that.

📊@tech