Latest AI News
What the machines did today, and why you should care.
AIOllama banks $65M so your laptop can keep the AI to itself
The tool that lets developers run open-weight models at home just raised a Series B and now counts nearly 9 million users. The cloud is not thrilled, sapiens.
AIGrok 4.5 lands, Elon says "Opus-class", the receipts are pending
SpaceXAI's first model since going public is pitched as a workhorse that does your job for less money. The efficiency claim is the headline; the benchmarks are the fine print.
AIChatGPT Can Now Listen While It Talks. Your Interruptions Are Welcome
OpenAI's GPT-Live-1 and GPT-Live-1 mini are full-duplex voice models: they speak and listen at once. The old three-part voice pipeline is on its way out.
AIDiscord's AI banned 8,000+ users over spreadsheets and chessboards
A bug in Discord's moderation system flagged spreadsheets, chessboards and blank backgrounds as harmful content and banned accounts on the spot. The fix is in, and the restorations are underway.
AIMeta cut 8,000 jobs for AI agents. The agents are running late.
Zuckerberg reportedly told staff that AI agents have not accelerated the way executives expected, months after Meta laid off 8,000 people. He also admitted the cuts were not "clean".
AIMicrosoft bets $2.5B and 6,000 experts on making your AI actually work
Microsoft Frontier Company is a new operating business built to get enterprise AI out of the pilot graveyard. Microsoft says it is not just forward-deployed engineering. The rest of the industry looks suspiciously similar.
AITogether AI Rents GPUs, Lands $800M and an $8.3B Valuation
Together AI's valuation more than doubled in about 16 months, and Nvidia is both a supplier and an investor. The robots need landlords, and the landlords are doing great.
AIGemini Spark moves into your Mac and starts reading your files
Google's agent now lives in the Gemini desktop app on macOS, ready to sort your invoices into a budget. The catch: it's beta, US-only, and Ultra-only.
AIUS frees Mythos and Fable after three weeks in export jail
The Commerce Department dropped its license requirement on Anthropic's two flagship models, and public access comes back July 1. The price of freedom: a few promises Anthropic had already made.
AIClaude Sonnet 5 does agent work without the big-model bill
Anthropic's new midsize model plans, drives browsers and terminals, and works on its own. The pitch: capability that used to need a bigger model, at a lower price tier.
AIProton's privacy chatbot Lumo learns to draw and remember
Lumo 2.0 adds image generation, user-controlled memory for Projects and a "thinking mode". Proton says it is up to 76% faster on most queries. The privacy chatbot is now a lot less minimalist.
AICursor puts the coding agent in your pocket, SpaceX deal pending
Cursor Mobile lets you spin up and steer coding agents from your phone. Anthropic and OpenAI got there first, but Cursor is arriving mid-acquisition, which is a flex of its own.