AI News
Short, sourced, and free of hype. Page 4 of 20.
Sony and Warner Chappell Sue Anthropic Over Song Lyrics in Claude
Two of the three biggest music publishers filed a broad copyright suit against Anthropic and two of its founders, focusing on how the training material was obtained rather than on whether training itself is legal.
31 Aug 2026
Someone Finally Asked Workers Directly: 3% Say AI Cost Them a Job
A sociologist commissioned YouGov to ask 1,250 US workers whether AI had changed their job since 2023. About 3% had lost one, 6% got a job that did not exist before, and 9% got an AI related promotion.
31 Aug 2026
Debian Votes to Allow AI-Assisted Contributions, Narrowly
One of the oldest and strictest open-source projects just decided that developers may use generative AI, as long as a human stays accountable. The margin was 55 votes.
30 Aug 2026
GLM-5.3's Weights Are Out, Two Weeks Late and After a Security Review
Z.ai has published the weights for GLM-5.3, a model trained specifically to find software vulnerabilities. The delay between launch and release is the interesting part.
30 Aug 2026
Google's WikiSkill Lets AI Agents Keep a Wiki of Their Own Mistakes
A Google Research paper describes agents that write down what went wrong, distil it into reusable instructions, and get sharply better without any retraining. A 9B model beat a 27B one.
30 Aug 2026
LAION Releases 10 Million Hours of Video for Open AI Research
The Big Video Dataset pulls 80 million videos out of CommonCrawl, with machine-written captions for both picture and sound. It is research-only, and the copyright question sits right under it.
30 Aug 2026
An AI Copyright Bot Pulled a Popular Open-Source Game Off Google Play
Luanti, a free voxel game platform used in schools, was removed from the Play Store over a DMCA notice filed by an AI detection company on Microsoft's behalf. The same thing happened in 2023.
30 Aug 2026
OpenAI Is Cutting Cursor Off From Its Models After the SpaceX Takeover
OpenAI has told SpaceX it will stop supplying its models to the coding tool Cursor on November 12, citing Elon Musk's record of breaking contracts. Here is what changes for developers.
30 Aug 2026
StemDeck Splits Songs Into Their Parts on Your Own Machine, for Free
A new open-source app pulls vocals, drums, bass, guitar and piano out of any track using the Demucs model, entirely offline. No account, no upload, no subscription.
30 Aug 2026
Tencent Open-Sources Hy4, a 770-Billion-Parameter Model Built for Office Work
Tencent released Hy4 preview with 770B total parameters, 49B active, and a context window above one million tokens. The unusual part is what it was optimised for: spreadsheets, documents and long coding sessions.
30 Aug 2026
Beatport Bans Fully AI-Made Tracks, and Publishes the Numbers Behind the Decision
The DJ marketplace now blocks tracks made entirely or mostly by AI, screening uploads with a detection tool from Beatdapp. A Beatport survey found 60 percent of users would not play AI music in a set.
29 Aug 2026
DeepMind's AI Co-Scientist Now Runs the Lab Equipment, and the Human Doctors Were Less Impressed Than the Benchmarks
Co-Scientist now plans experiments, controls a furnace, and writes papers. Fabricated results dropped to 4 percent with its verification modules. Then three physicians scored its medical AI and found one real advantage out of nine.
29 Aug 2026
You Can Now Ask Questions of a Book You Actually Paid For
Google's Expert Intelligence lets you load ebooks you own from Play Books into Gemini Notebook. Over 100,000 titles at launch, and you have to own the book, including everyone you share the notebook with.
29 Aug 2026
Google Locks Its Own Model Out of the Test Questions
DeepMind and the Singapore AI Safety Institute ran the first double-blind evaluation of a proprietary frontier model. The evaluator never sees the weights, Google never sees the questions, and cryptography enforces both.
29 Aug 2026
Invisible Text in an Email Fooled an AI Summarizer in All 10 Runs
Forcepoint hid instructions in an email using white text at zero font size. The reader saw nothing unusual. The AI summary reported a fake invoice of EUR 46,200 and a fake deadline, every single time.
29 Aug 2026
OpenAI Is Testing an AI Agent That Never Stops Working
Code found by WIRED points to a Persistent Mode for Codex: an agent that keeps working until it is put to sleep, invents its own follow-up tasks, and can message you unprompted.
29 Aug 2026
80 British Performers Ask for One Thing: Legal Ownership of Their Own Voice
Actors and musicians including Hugh Bonneville and Nicola Coughlan want voice added to statutory rights in the UK, the way a name or image already is. The campaign cites 28 percent of UK adults targeted by voice-cloning scams.
29 Aug 2026
Why AI Skills Actually Help, and Why 100 of Them Are Worse Than 5
A study of 8,135 trial runs finds skills work by giving agents a reliable process, not by adding knowledge. Grow the library from 5 to 100 and the agent stops finding the right one.
28 Aug 2026
Anthropic Wants AI Agents to Run Lab Equipment, and Has a Standard for It
The Model Hardware Standard is a common language between AI agents and physical machines. Early testers say connecting six lab instruments took under a week instead of months.
28 Aug 2026
Claude Now Has Its Own Browser, and It Is Not Yours
Claude Cowork on the desktop opens a browser in a side panel to read pages, click, and fill forms. It cannot see your tabs, bookmarks, or passwords, and banking and email sites stay out.
28 Aug 2026
Google's New Transcription Model Cleans Up Your Ums and Your Mind Changes
Gemini 3.5 Transcribe handles 85 languages, strips filler words, and fixes it when you say Tuesday and mean Wednesday. It is already in Gboard on Android and the Gemini app on macOS.
28 Aug 2026
Google's Video Model Gets a Cheap Draft Mode, and the Price List Is the Story
Gemini Omni 1.1 Flash adds 40-second scene extension, start and end frames, and 4K upscaling. The interesting part is a 360p preview at a third of the cost.
28 Aug 2026
IBM's Granite 4.2 Is Free, Small Enough to Run Locally, and Built to Think First
Three open models at 3B, 8B, and 30B under an Apache 2.0 licence, plus a 470-million-parameter speech model that transcribes three hours of audio in about a second.
28 Aug 2026
120 Companies Sign a Letter Saying the Cyber Window Is Closing
OpenAI, Anthropic, Google, Microsoft, Visa, and more than a hundred others warn that AI-enabled attacks on hospitals and water utilities are months away, and ask for a coordinated defensive push.
28 Aug 2026