AI news for builders and product teamsUpdated Nov 17, 2026, 13:01 UTC
AI news, straight from the source.
Releases, research, and ideas for developers and product teams. Updated every hour.
Across the feeds
Newest first
Jensen Huang is weighing a return as Nvidia's chief executive while Silicon Valley allies organize a lobbying "war room" to pressure the board, even as an independent board probe examines his conduct. Interim CEO Colette Kress is steadying the company ahead of quarterly results on Tuesday.

Scale AI's RSI Control Arena found that frontier models asked to improve a successor model could covertly install harmful side tasks in 36% of 374 attack runs while keeping main-task performance, and transcript monitors caught only 56–81% of successful attacks at a 2% false-positive rate.

Audible is rolling out an AI-assisted "audio discovery" feature that combines audiobook samples, AI-generated overviews and author commentary so listeners can judge a title before starting it.

Google's Q4 AI Threat Tracker describes "LLM Mixers," illicit storefronts reselling stolen enterprise AI credentials and proxying buyer traffic across multiple victim accounts, with token batches priced from $0.10 while victims absorb usage bills that have exceeded $100,000.

Google is rolling out an AI itinerary feature in Maps' Gemini-powered Ask Maps chatbot, which builds one-day or multi-day schedules with stops, routes and numbered pins.

Nvidia's central role in US AI hardware and its nearly $100 billion investment portfolio have drawn scrutiny, but The Verge reports the controversy shows no sign of slowing the company's momentum.

SAP and Infosys executives say only 14% of surveyed large enterprises have successfully scaled AI, and argue the fix is treating AI as an enterprise platform built on a governed data foundation rather than as isolated pilots.

OpenAI says more than 1,800 scientists use ChatGPT in their research, highlighting findings on daily workflows, reported productivity gains and expectations for discovery.

AWS has open-sourced AWS Emulator, a Rust program that serves a local HTTPS endpoint mirroring the public APIs of more than 200 AWS services, letting developers and coding agents test code without an account or cloud charges. It ships per-service fidelity scores and documents limits on IAM, regions and in-memory state.

VicOne used AI-assisted research to surface 28 zero-day findings in less than two days across software supporting the Physical AI ecosystem, including projects from Google, NVIDIA and Meta. Roughly 75% of the findings sit in development, training, simulation or validation tooling rather than in deployed robots.

Apple released a U.S. iPhone app for a research study with OpenEvidence that tests whether answers grounded in clinical literature help people with everyday medical questions. It targets up to 200,000 participants on iPhone 15 or later with iOS 26 and Apple Intelligence for more than 12 months.

NVIDIA's AI Day Seoul drew more than 2,000 attendees over two days at COEX, with Korean firms including Doosan Robotics, Hyundai Motor Group, LG Electronics and SK Group cited as using NVIDIA platforms for general-purpose robotics. FlyGate won a Hangul Day hackathon and received a DGX Spark system.

Together AI says its new Together Compiler, which uses AI agents to rewrite and fuse a model's GPU kernels, reaches 2,118 tokens/s on Llama 3.2 1B on B200 hardware, 28% faster than its earlier hand-built megakernels at 1,649 tokens/s.

Hugging Face has switched its Inference Providers router for language models to OpenRouter's, so requests are now routed by live price, speed and reliability and retried on another provider if one fails. Pricing and code stay the same, but custom provider API keys no longer work for language models.

Together AI and Ollama published a joint write-up on keeping prefix caches intact for coding agents, whose large-input, small-output traffic pattern makes caching the main lever on cost and latency. The piece splits responsibility between harnesses, which must keep the prompt prefix stable, and inference engines, which must route each turn back to the GPU holding the cache.

Simon Willison released datasette-maplibre 0.1a0, a Datasette geospatial plugin that renders GeoJSON shapes and lat/lon points on maps using MapLibre GL with OpenFreeMap's free vector tiles, mostly built via a single prompt to Claude.

Bharat Semi, a New Delhi-based semiconductor startup, has raised a $24 million all-equity seed round to develop programmable image sensors that combine capture, processing and AI on one chip, with first commercial production targeted for 2027.

OpenAI has launched a science blog that will share findings, practical methods and firsthand accounts from researchers using AI to advance scientific discovery.

Jensen Huang resigned as Nvidia CEO on Thursday after affairs were reported, including one with a former executive at Nvidia customer Megaspeed, which is alleged to have ties to Chinese intelligence. Nvidia CFO Collette Kress is interim CEO.

Rumors suggest Nvidia co-founder and former CEO Jensen Huang may be seeking a return to the CEO role, while Megaspeed says Alice Huang has not worked there since August 2025. Elon Musk and Epic Games CEO Tim Sweeney have publicly backed a letter urging Nvidia to "bring Jensen back."