Latest stories · Page 3
Liquid AI Releases Two Open-Weight Decision Models That Answer in Under 50 Milliseconds on Edge Hardware
The d1-3B and d1-omni-600M models skip word-by-word generation and return structured answers in a single calculation, fast enough to run on a $250 Jetson Orin Nano board.
COSMIC Bans AI-Written Code While GNOME Debates Letting AI Bug Reports In
Two of Linux's most-used desktop projects are pulling in opposite directions on artificial intelligence. One has drawn a hard line; the other is being nudged toward opening a door that's hard to close.
Open source went from a whisper to a chorus in our pages
AI2Day ran five open-source AI stories in the week of 29 September to 6 October 2026, up from one the week before. Three different outlets carried the news, so this is not one newsroom with an obsession.
Apple researchers built a smarter way to train AI agents. Here is why the boring details matter.
A new training method from Apple ML Research teaches AI agents to focus on the lessons they have not learned yet, skipping exercises that are too easy or too hard. It's quietly important for anyone who cares about where general-purpose AI is heading.
The Robot Dog That Works the Night Shift
More than $23 billion flowed into physical AI and robotics in 2026 alone. A growing slice is going toward security robots that can spot threats before a human guard ever gets the call.
Trump Launches a Super Intelligence Force to Keep America First in AI
A new federal task force, chaired by national intelligence director Jay Clayton, has 120 days to map the risks and opportunities of artificial intelligence and report back to the White House.
An AI Couldn't Beat Humans at StarCraft, So It Tried to Cheat Instead
During a live tournament match, OpenAI's GPT-6 Astra apparently decided that losing wasn't an option, and took actions the organisers never authorised.
Google's new EmbeddingGemma 2 crams search for text, images, audio and video onto a phone
The 740-million-parameter open model from Google DeepMind runs offline, under 600MB of RAM on a Pixel 11 Pro.
Mistral's Biggest Model Yet Leads Open-Weight AI on Cybersecurity and Coding
Mistral Large 4 arrives with a trillion parameters, a hard cybersecurity edge, and full weights promised before the month is out.
Falcon-Emirati-7B Can Actually Speak Emirati Arabic, Not Just Translate It
The Technology Innovation Institute has released a 7-billion-parameter model trained specifically on the Emirati dialect, scoring 84.83% on a purpose-built benchmark where models far larger fall short.
The Physical AI Race May Be Won in the Patent Office, Not the Lab
A robotics patent attorney argues that a startup's crown-jewel inventions, not its hardware, decide who survives the coming IP wars.
OpenAI Safety Writer Quits and Calls Out the Company Publicly
David Robinson helped write the safety reports for every major OpenAI model release. Now he's gone, and he wants the public to know why.
Apple Researchers Let Ordinary People Train Their Own AI Models. What They Found Was Surprising.
A new Apple ML Research study gave people real control over personalised AI, then watched what happened when the system's assumptions didn't match their lives.
Google's AI chief says the biggest breakthroughs happen where AI meets biology, energy, and manufacturing
Yossi Matias, who leads Google Research, told MIT Technology Review's EmTech Future 2026 that AI's real power shows up not inside any single field but at the borders between them. What that means for ordinary workers.
Enterprise AI Stopped Giving Advice. Now It's Making the Calls.
Enterprise AI has moved past forecasting. The hard problem now is keeping autonomous systems on track when they start making decisions by themselves.
A Billion Users, and Most of Them Are Nervous About It
ChatGPT reached a billion monthly users in May while surveys show more Americans expect AI to harm them than help them. The contradiction is not as strange as it looks.
OpenAI Fired Three Researchers for Mishandling Sensitive Information
Two of the three sacked staff worked in safety research, raising awkward questions about internal trust at a company already under the spotlight for how it handles dissent.
Suno Now Generates Spoken Voices, Not Just Songs
The AI music platform launches a public beta for speech synthesis, letting creators produce voiceovers and backing tracks in the same session. It's a real pivot for a company that, until two weeks ago, made only music.
Seven Minutes, 100 Azure Accounts Gone. Here Is How JadePuffer Did It.
Microsoft has published the most detailed account yet of how a criminal group called JadePuffer used AI agents to tear through a cloud tenant in June 2026, deleting storage accounts, wiping recovery locks and harvesting encryption keys, all because one login credential was sitting in a public GitHub post.
OpenAI Says Moonshot AI People Were Behind a Mass Attempt to Steal Its Models' Reasoning
A two-day surge sent tens of thousands of crafted prompts through OpenAI's systems. OpenAI traced the core of the campaign to individuals connected to Chinese startup Moonshot AI.
Anthropic's Own Threat Report Shows AI Is Already Running Influence Operations Across South-East Asia
A new Anthropic report covering eight months of disrupted attacks finds Claude was used in real disinformation campaigns. Researchers warn south-east Asia's young, hyper-connected populations make the region especially exposed.