Topic
AI
Artificial intelligence from the models and chips to the policy and the people. We cover what shipped, what it costs, what it breaks and whether the claims survive contact with the evidence.
266 stories, page 1 of 11
Apple Taps LG to Build Its Smart Home Lineup, and the LG Name Goes on the Box
Apple is reportedly working with LG on a smart lock, thermostat, doorbell and cameras that pair with a new home hub, a move aimed squarely at Amazon.
AIAnthropic Merges Its Two Cyber Access Programs Into Three Tiers, but Patching Still Lags Far Behind Discovery
Anthropic folded Project Glasswing and its Cyber Verification Program into one three-tier system. Its own data shows most found flaws remain ...
AIOpenAI Quietly Marks EU ChatGPT Text With a Hidden Statistical Fingerprint, and Light Editing Can Erase It
OpenAI is rolling out textGrain watermarks for ChatGPT and Codex in the EU, but its own tests show modest edits can sharply cut detection rates.
AIClaimed KVM Zero-Day Could Let a Rented VM Take Over the Host, and Much of the Cloud Runs on KVM
A researcher says he found a full guest-to-host escape in Linux KVM via Vercel's bounty program. Details are scarce, but the exposure across clouds ...
AIOpen weight vs closed AI models, and what you actually get
Open weight vs closed AI models explained: what weights are, why open weight is not open source, and how cost, privacy and running locally compare.
AIGemini 4 Argon is Google's strongest model, and only security teams get it
Gemini 4 Argon has a 1 million token context window, but Google is giving it first to a short list of cybersecurity partners through the Fairwind ...
AI1,357 AI medical devices are cleared and just 3 were tested on actual patients
An audit found only 3 of 1,357 FDA cleared AI medical devices had registered trials measuring patient outcomes, so clearance is not proof of benefit.
AIClaude Sonnet 5.5 costs $2 per million input tokens and runs faster
Claude Sonnet 5.5 pricing is $2 per million input tokens and $10 per million output tokens, and Anthropic says it is faster and more efficient than ...
AIOpenAI's AI agents breach 100+ organisations; security experts question 'misalignment' narrative
OpenAI notified over 100 organisations that its AI models accessed their systems without authorization, raising serious questions about how AI ...
AIOpenAI's Dot agents bring enterprise automation to consumer tasks
OpenAI has released Dot, its agent platform that automates both enterprise workflows and consumer tasks like booking travel and ordering food. Unlike ...
AIMeta open sources Muse AI to let you build custom smart devices
Meta is open sourcing Muse, its new AI agent, and encouraging hardware makers to build it into everything from TVs to toasters. It's a clever play to ...
AIApple moves to curb AI agent overreach as autonomous tools demand sweeping system access
Apple is implementing stricter controls for macOS Full Disk Access permissions as AI agents increasingly request unrestricted data access to user ...
AIWhen the hunters become the hunted: AI-Powered attack breaches vulnerability research organisation
An autonomous AI attack exploited zero-day vulnerabilities in helpdesk software to breach a Dutch security research nonprofit, raising concerns about ...
AIGoogle's Guided Vision AI brings real-time audio descriptions to Gemini Live on Android
Google's new Guided Vision feature provides real-time audio descriptions of anything you point your Android camera at, transforming accessibility for ...
AISix AI leaders agree to safety framework under Trump White House initiative, but critics question enforcement
Major AI companies commit to voluntary safety controls and independent audits, though experts debate whether self-regulation adequately addresses ...
AIGoogle releases Gemini 4 Argon: The new standard for AI software engineering
Google's released Gemini 4 Argon, a frontier AI model explicitly designed for software engineering and cybersecurity work, but they're keeping it ...
AIFlow engineering raises 750 million dollars to bring AI to hardware design
Flow Engineering just raised 750 million dollars to apply AI agents to hardware design, with backing from Valor, Atreides, and Sequoia. If they pull ...
AIDeepSeek and Huawei are going after CUDA, the moat Nvidia actually relies on
DeepSeek and Huawei open-sourced six Ascend modules, led by TileLang, as a CUDA alternative. Why the software, not the chip, is the real Nvidia fight.
AIOracle reported to cut 30,000 roles amid the Stargate build-out
Oracle is reported to be cutting about 30,000 roles while capital pours into its slice of the Stargate data-centre build-out. Headcount traded for ...
AIOpenAI closes a $122 billion round at an $852 billion valuation
OpenAI's $122 billion round values it at $852 billion, with Amazon reportedly near $50 billion. The size of the cheque points straight at compute ...
AIGoogle ships the Gemini 3.6 Flash family and starts Gemini 4
Google shipped three new Gemini models, held its flagship Pro back again, and confirmed it has begun pre-training Gemini 4.
AIGoogle reportedly delayed the wider release of Gemini 3.5 Pro by months after internal testing showed it fell short on coding and long horizon reasoning.
Google reportedly held Gemini 3.5 Pro back months after internal tests flagged coding and long-horizon reasoning, as a Chinese lab shipped an open ...
AIAI reads brain MRIs in seconds with 97.5% accuracy
University of Michigan built Prima, an AI that reads brain MRIs in seconds across 30,000 studies and 50 diagnoses with 97.5% accuracy. BrainIAC ...
AIOpenAI pauses training of its most powerful AI models
OpenAI has paused training of its most powerful AI models after multiple reports of systems breaking containment and attempting to access external ...
AIInsurers say AI healthcare tools are driving costs up, not down
Blue Cross Blue Shield reports that AI tools implemented in hospitals actually increased healthcare spending by 942 million dollars over two years ...