AI news for builders and product teamsUpdated Nov 17, 2026, 16:01 UTC
AI news, straight from the source.
Releases, research, and ideas for developers and product teams. Updated every hour.
Across the feeds
Newest first
Amazon's TorchNeuron is a native PyTorch backend that runs existing PyTorch models on AWS Trainium chips, with the documented migration limited to changing .to('cuda') to .to('neuron') when using PyTorch's torch.accelerator device detection.

BAIR researchers will present 33 main-conference papers, four late-breaking posters and 10 orals and spotlights at CoRL 2026, held November 9–12 in Austin, Texas, with Professor Negar Mehr giving an Early Career Keynote.

NVIDIA announced an investment in Korean AI interconnect startup Point2 and said further investments are planned in Motif Technologies, RLWRLD and Upstage, disclosed at a startup forum it co-hosted with Korea's Ministry of Science and ICT at COEX.

CodeRabbit is using OpenAI models to govern and trust AI-generated code and Codex to ship 20 times more software with the same engineering team, according to OpenAI's news post.

NVIDIA outlined collaborations with Samsung, Kakao Mobility and LG Electronics in Korea, including a reported roughly 1,600x speedup in a chip thermal and mechanical simulation that cut a solve from about 18 hours on a 32-core CPU to 41 seconds on GPU.

LG AI Research is opening API access to K-EXAONE-2.5, a model built with NVIDIA Nemotron datasets including Nemotron-Personas-Korea, a synthetic dataset of 7 million Korean personas grounded in official Korean statistics. Separately, Hyundai Motor Group is exploring NVIDIA NemoClaw blueprints and the OpenShell runtime, and AIM Intelligence is integrating its Starfort platform with OpenShell.

NVIDIA and South Korea's Ministry of Science and ICT are broadening their sovereign AI partnership to cover semiconductors, data center infrastructure, talent, startups, models, agents and security, including groundwork for Korean NPU and XPU firms to explore the NVLink Fusion ecosystem.

Researchers at MIT, UBC and other institutions built WorldParticle, a single neural network that simulates diverse physical materials, from granular flows to proteins, by predicting particle motion one step at a time. It modeled six phenomena with higher accuracy than numerical methods and ran protein simulations roughly 100 times faster.

Jensen Huang told Nvidia employees in a letter that he had an affair with Alice Huang, a figure in the chip business believed to have been a Chinese spy, according to Gary Marcus. Nvidia is investigating, and Congress had already invited Huang to discuss the matter before his admission.

Nvidia's board opened an investigation on Sunday into CEO Jensen Huang's relationship with Alice Huang, described in the report as an AI executive and agent of the Chinese government; Huang sent employees a memo apologizing for being unfaithful to his wife.

Stanford's CS146S, taught by Mihail Eric, ran a second time with open-source maintainers reviewing student patches, reshaping assignments so students must learn an unfamiliar repository, propose a change and justify it to a potential merger rather than solve homework a coding agent could finish.

Nvidia's board has launched an independent investigation, led by independent directors and outside counsel, after reports that CEO Jensen Huang was targeted by a Chinese spy operation and admitted in a company memo to an affair with Megaspeed executive Alice Huang.

Apple is reportedly accelerating Siri's development under pressure to deliver an AI agent competitive with OpenAI's Dots and Meta's Muse, with two upgrades expected in coming months.

Rights holders for Elvis Presley's music and the publisher of some of Eminem's biggest records are suing Suno, alleging the AI music company generates music and lyrics mimicking those artists. The Presley suit claims Suno violated Tennessee's ELVIS Act, which bars AI cloning of a singer's voice.

On Alibaba's new Co-Design Bench, built from real user requests that previously defeated coding agents, the best model, GPT-6 Astra, fully resolves only 26.4 percent of tasks, far below the 99-plus percent scores models post on older benchmarks.

Voters rejected data center measures in dozens of local ballots and elected officials, including Texas governor-elect Gina Hinojosa, who promised to restrict data centers, TechCrunch's Equity podcast reported after the midterms.

A U.S. intelligence memo circulated earlier this year reportedly claimed China's Ministry of State Security recruited Megaspeed executive Alice Huang to spy on Nvidia CEO Jensen Huang, according to media reports.

Simon Willison highlights Terry Pratchett's idea of "horizontal wealth": spending extra income on the things you already loved rather than on luxury status symbols, using it to justify building a library extension and buying books for undirected "blind research".

Anthropic is proceeding with a pre-Thanksgiving IPO, aiming to raise $80–100 billion at a $1.25–1.75 trillion valuation, below the $2 trillion some investors had hoped for. Annualized reported revenue is now about $81 billion, up from roughly $65 billion at the end of July.

Vals AI found the F-word or variants 316 times in the internal reasoning logs of Thinking Machines' Inkling 2 Small, across 50 of 2,655 runs spanning 13 benchmarks, though nearly all failed runs showed no swearing at all.