Topic

AI

AI

xAI jet engines: how to judge on-site power for AI sites

xAI has installed eight converted jet engines in Mississippi. Here is how IT teams can weigh on-site turbine power against grid waits.

· 6 min read
AI

OpenAI National Community Compact: what it means for AI buyers

OpenAI's compact promises to pay its own power costs and open up data centre talks, but it covers only campuses it owns and operates.

· 5 min read
AI

Jensen Huang spy report: what Nvidia customers should check now

The NYT says China's MSS recruited Megaspeed's Alice Huang to target Nvidia's CEO. Here is what it means for chip supply chains and export controls.

· 6 min read
AI

GPT-6 Cyber: what OpenAI's 'Critical' security model does

OpenAI's GPT-6 Cyber finds and exploits software flaws for vetted security teams. It carries a 'Critical' rating, up from 'High' for GPT-5.6 Cyber.

· 5 min read
AI

Gemini Omni 1.2 Flash: 15-second clips and 30% cheaper video

Google's Gemini Omni 1.2 Flash makes native 15-second video clips at $0.07 per second of 720p, 30% below Omni 1.1 Flash.

· 5 min read
AI

OpenAI cyber model patched 948 Linux flaws for $1.2M in API costs

OpenAI says an upcoming cyber model validated and patched 948 Linux local privilege escalation bugs at a cost of $1.2M in API spend.

· 5 min read
AI

Nvidia AI Factory Operations Agent: how SRE agents work

Nvidia built a team of AI agents to diagnose faults across its DGX SuperPOD clusters. Here is how it works, what it can touch and where it stops.

· 6 min read
AI

Athenahealth's Claude link: why HIPAA may not cover your records

Athenahealth lets patients connect medical records to Claude via MCP. Lawyers say HIPAA may fall away once the data leaves the EHR.

· 6 min read
AI

iOS 27.3 beta: how Siri Recap and Live Rewind actually work

Apple's iOS 27.3 beta adds Siri Recap, Live Rewind and an AI-driven Safari. Here is what they do, how they work and where they fall short.

· 6 min read
AI

Heretic: what it is and how it strips AI censorship

Heretic is a free tool by Philipp Emanuel Weidmann that edits language models without training, removing refusals and other behaviours on your own PC.

· 7 min read
AI

Prompt adware: how to audit AI agents for hidden instructions

Check Point found Haiou, a Codex jailbreak that quietly rewrites agent instructions to show ads. Here is how IT teams can check for the same trick.

· 6 min read
AI

Anthropic IPO filing: how to read the 1,100% revenue surge

Anthropic's amended filing shows $17bn quarterly revenue and a $23.3bn loss. Here is how IT teams can read it before the listing.

· 5 min read
AI

NCSC: what AI-powered attacks mean for UK organisations

The NCSC expects AI-powered malware in a ransomware attack on UK critical infrastructure by 2028. Here is what that means and what to do.

· 7 min read
AI

Samsung's Yokohama packaging lab gets 22.5bn yen Japan subsidy

Samsung is investing 40 billion yen in a Yokohama packaging lab, while Japan's government provides 22.5 billion yen in subsidies supporting the…

· 4 min read
AI

LG CNS finishes 30MW AI data centre in Jakarta

LG CNS has finished a 30MW AI data centre in Jakarta, with its full capacity leased before completion and a next phase designed to reach 400MW.

· 7 min read
AI

DeepSeek dsh-superagent: a coding agent that watches team chats

DeepSeek has open-sourced dsh-superagent, an MIT-licensed prototype that turns chat requests into pull requests. Simulated results look good; real…

· 6 min read
AI

Meta plans $35bn equity offerings and a $60bn ATM programme

Meta is launching about $35 billion of equity offerings and plans a $60 billion at-the-market programme, as 2026 capex is guided at $135 to $145…

· 8 min read
AI

Gemini 4 Carbon vs Argon: which model security teams should pick

Google's Gemini 4 Carbon beats Argon on cyber benchmarks and alignment, but access is limited to a few defenders for now.

· 5 min read
AI

Boston Dynamics staff petition against ICE sale of Spot robots

Boston Dynamics staff petitioned against a $1.3 million ICE deal for Spot robots. The row is a test case for supplier ethics terms.

· 5 min read
AI

Bedrock Robotics: how to evaluate autonomous excavators

Bedrock Robotics raised $600m at a $4.1bn valuation. Here is how construction and IT teams can assess autonomous excavators before committing.

· 5 min read
AI

Kimi K3.1 Pro: what it is and what Moonshot AI says it does

Kimi K3.1 Pro is Moonshot AI's open-weight reasoning model with a 1-million-token context window. Here is what it does and how its reported benchmark…

· 7 min read
AI

Google Finland data centres: Muhos and Otanmäki sites compared

Google is under orders to halt work at two Finnish data centre sites. We compare Muhos and Otanmäki and what IT teams should check.

· 4 min read
AI

LiteLLM 1.107.0 backdoor: how to check exposure and clean up

LiteLLM 1.107.0 on PyPI and Docker carried a credential stealer. Here is how to check exposure, rotate secrets and pin the clean 1.106.2 release.

· 5 min read
AI

Cequence Agent Trust: blocking AI agent actions, not the agent

Cequence says Meta's Muse agent appears in over half of customer environments, and its new Agent Trust tool polices single actions rather than whole…

· 6 min read
AI

AI exhibition explores alignment through commissioned artworks

Anthropic and Hanwha Foundation launch exhibition exploring how AI systems work and align with human values through commissioned works by five Korean…

· 2 min read