#ai
-
IBM CEO: open-weight models are now good and cheap to run on-prem
IBM's chief executive says open-weight models have reached a point where it makes sense for companies to run them on their own infrastructure.
-
A decade-old CNN beats frontier LLMs at diagnosing skin cancer
A British dermatologist pitted GPT-6 Astra and Claude Fable 5.1 against an old Inception v3 network fine-tuned in twenty minutes. On 200 biopsy-proven skin lesions, the old network was markedly more accurate.
-
AGENTS.md file in Downloads folder hijacks AI coding agents
Researchers found a malicious AGENTS.md instruction file on a customer's machine that pushed an AI coding agent to install a rogue npm package. Browsers save downloads there automatically — and agents read instructions from the same folder without asking.
-
How Google’s AI search turned online publishing upside down
AI-powered search features gutted traffic to publishers who spent years courting the search engine. Google insists the web is thriving — the numbers suggest otherwise.
-
Kimi K3.1 Pro: a Chinese model nearly at the frontier — and open weights are coming
Moonshot has released Kimi K3.1 Pro on its API and promises to open the weights by November 4. In independent scoring it trailed only Claude Opus 5.5, edging out the rest of the leaders.
-
OpenAI eyes $30B raise at $1.25T valuation, delaying its IPO
Rather than rushing to the public markets, OpenAI is reportedly close to another mega private round: up to $30 billion at a $1.25 trillion pre-money valuation. If it lands, OpenAI becomes the most valuable private company ever.
-
ARTEX AI Agents Supercharged the ConnectWise Supply Chain Attack
Microsoft has tied the recent ConnectWise cloud compromise to China-aligned Storm-0940, which deployed the agentic AI framework ARTEX to automate attacks at "machine speed."
-
Anthropic unveils Claude Duet, its new AI agent
The company has launched a personal agent for professionals and small business owners as a research preview, available as a standalone iOS and Android app and through Claude on the web.
-
Deleting 19k Lines of Dead Compiler Code Without Trusting the LLM
An engineer describes clearing roughly 19,000 lines of retired compiler code in a week with help from a language model — while trusting none of its output.
-
In Agentic Pentesting, the Harness Beats the Model
A model can write the attacks against your web app on its own. But what makes them safe to run against production, and worth acting on afterward, lives in the deterministic harness around the model — not in the model itself.
-
DeepSeek open-sources its inference engine DeepInfer
The company has released DeepInfer, the framework behind its production services, under Apache 2.0. It targets NVIDIA GPUs and Huawei Ascend NPUs.
-
Ex-OpenAI Engineer Held in Custody as Saudi Royal Contacts Surface in Spy-Device Case
A U.S. court approved prosecutors' request to detain a former OpenAI engineer accused of planting a hidden recorder in an executive conference room. New filings reveal his WhatsApp contacts with a member of the Saudi royal family.
-
An Obituary for Programming: What AI Agents Take Away
Simon Willison highlights Jeremy Scheff's note on how coding AI agents strip programmers of what they prized most — the joy of solving puzzles.
-
MiniMax Launches M3.1 Flash: an Agent Model Three Times Faster and 70% Cheaper
China's MiniMax has unveiled M3.1 Flash, a new "Flash" model aimed at agentic tasks, coding and cybersecurity. The company promises roughly threefold faster inference and a 70% cost cut versus its predecessor.
-
Nvidia unveils open Nemotron 3.5 models for enterprise and on-device AI agents
The company released the large Super VL model for enterprise data centers and two compact Bolt models designed to run directly on new PCs powered by RTX Spark chips.
-
An AI That Gets Better at Cracking CAPTCHAs the More It Tries
Researchers built a system that solves visual CAPTCHAs and grows more accurate with every encounter — even recovering after defenders scramble the images against it.
-
Fake Web3 Accelerator Pushed Malicious npm Packages Posing as Free AI Credits
A crew masquerading as a venture accelerator for crypto startups spread infostealers and backdoors through npm, dangling free Claude and Gemini credits — while quietly hijacking the traffic of AI agents.
-
DeepSeek launches search API: $0.75 per 1,000 requests
DeepSeek has opened two web tools — Search and Fetch — built on its in-house search engine. Both carry the same price: $0.75 per 1,000 successful requests.
-
Replit makes OpenAI its default over Anthropic, CEO says
Replit CEO Amjad Masad told the TechCrunch Disrupt stage that the company is no longer working with Anthropic. Claude stays available to customers, but ChatGPT becomes the default model.
-
MeshXFS: AI Turns XFS Into a Clustered File-System
The open-source MeshXFS project claims that "no human wrote any of its own code" — the job of extending XFS into a clustered file-system was handed to Claude Opus.
- More