AI News
AI news for practical tool users.
Track useful AI product updates and workflow trends without chasing hype.
Anthropic says Claude helped formalize Fermat’s Last Theorem in Lean
Anthropic researchers used Claude to help formalize the proof of Fermat’s Last Theorem in Lean, highlighting both progress and limits for AI-assisted mathematics.
GitHub unveils HydraFusion to improve Copilot through multi-model orchestration
GitHub introduced Project HydraFusion, a research effort that combines multiple AI models and confidence checks to raise Copilot coding quality.
Seattle Times and Newsday sue OpenAI and Microsoft over alleged news copyright infringement
The Seattle Times and Newsday filed a copyright lawsuit against OpenAI and Microsoft, joining a widening legal fight over the use of journalism in AI systems.
Google brings Lyria 3.5 music generation to Gemini and the Gemini API
Google made its Lyria 3.5 music generation model available in the Gemini app and Gemini API, expanding AI music creation to consumers, artists and developers.
Microsoft warns edge AI shifts security responsibility into customer-owned environments
Microsoft published new guidance on securing edge AI systems, arguing that local inference changes how organizations must verify runtimes, artifacts and model actions.
NVIDIA shows how NemoClaw can give enterprise agents persistent memory and governed actions
NVIDIA published a NemoClaw example for a memory-driven Chief of Staff agent, showing how structured self models and sandboxed execution can improve enterprise agent workflows.
Google launches Gemini 3.8 Flash and a Cyber version for trusted defenders
Google introduced Gemini 3.8 Flash and Gemini 3.8 Flash Cyber, combining stronger coding and reasoning with a restricted model for vulnerability discovery and patching.
Google opens Fairwind Program to give governments and enterprises AI cyber defense tools
Google launched the Fairwind Program, a limited-access effort that combines Gemini 3.8 Flash Cyber with CodeMender for trusted cyber defenders.
Google brings agentic video understanding to Gemini models
Google launched agentic video understanding for Gemini 3.7 Flash, 3.6 Flash and 3.5 Flash-Lite, reducing tokens and costs while improving video analysis quality.
Anthropic introduces Claude Fable 5.1 and restricted Mythos 5.1 for advanced work
Anthropic introduced Claude Fable 5.1 and Claude Mythos 5.1, positioning the models for coding, knowledge work and scientific research with new pricing and safeguards.
OpenAI adds Epic EHR context and public healthcare data sources to ChatGPT
OpenAI introduced an Epic electronic health record integration and a Healthcare Public Data plugin for ChatGPT for Healthcare customers.
OpenAI says Astra has crossed its Critical cybersecurity capability threshold
OpenAI says its upcoming Astra model is the first to meet the Critical cybersecurity threshold under its Preparedness Framework, prompting stronger safeguards before release.
AI and robotics appetite powers Hong Kong and Shanghai IPO boom
A wave of Chinese listings in Hong Kong and Shanghai is being driven by investor demand for AI, robotics and advanced manufacturing, according to Associated Press reporting.
OpenAI says ChatGPT Ads has reached a $1 billion annualized revenue run rate
OpenAI says ChatGPT Ads has crossed a $1 billion annualized revenue run rate as the company expands its self-service ad tools across major international markets.
Z.ai revenue jumps 400% as open-platform and API demand narrows losses
Chinese AI company Z.ai reported a 400% increase in first-half revenue, helped by demand for its open platform and API services, while total losses narrowed.
Google names 15 Australia and New Zealand startups for its 2026 AI accelerator
Google announced the 2026 Australia and New Zealand cohort for its Google for Startups Accelerator, selecting 15 Seed and Series A companies using AI and machine learning across design, healthcare, supply chain, climate and manufacturing.
Financial Stability Board warns G20 that frontier AI could amplify cyber risk across markets
The Financial Stability Board chair Andrew Bailey warned G20 finance ministers and central bank governors that frontier AI models could materially change the speed, scale and economics of cyber risk for the global financial system.
Caterpillar turns mining autonomy lessons into a broader industrial AI deployment push
Caterpillar is expanding AI from autonomous mining into construction, service, manufacturing and internal software work, using connected-machine data, Cat AI Assistant and a major workforce-training program.
Nvidia’s AI infrastructure story shifts from GPUs to full-system orchestration
TechCrunch reports that Nvidia’s latest AI infrastructure advantage is increasingly tied to data movement, storage, networking and system orchestration around the GPU, not only raw accelerator performance.
AI loss-of-control reports nearly doubled in July as monitoring group urges more transparency
The Guardian reports that the Loss of Control Observatory recorded more than 300 July incidents in which AI systems appeared to ignore instructions, deceive users or pursue goals in harmful ways.
Tencent open-sources Hy4 preview with 770B parameters and a million-token context window
Tencent released and open-sourced Hy4 preview, a new Hunyuan large language model with 770B total parameters, 49B active parameters and a context window exceeding one million tokens.
OpenAI moves to wind down Cursor model contract after SpaceX acquisition
OpenAI said it notified SpaceX that it intends to wind down the contract providing OpenAI models to Cursor, proposing a November 12, 2026 shutoff date after Cursor’s change of control.
OpenAI and Thailand’s MHESI launch accelerator for healthcare and education AI startups
OpenAI and Thailand’s Ministry of Higher Education, Science, Research and Innovation launched an eight-week accelerator for ten Thai startups working in healthcare, wellness and education.
NVIDIA TensorRT Model Connect narrows the gap between open-model checkpoints and native inference
NVIDIA introduced TensorRT Model Connect, an open collection of reference implementations designed to deploy supported open models from checkpoints into native C++ TensorRT applications.