Craig Hale

America's three biggest mobile networks reveal plans to bring direct-to-device satellite connections to customers.

Codex still runs on your desktop, but OpenAI is turning your ChatGPT mobile app into a remote controller for access from anywhere.

Claude Code has proven a hit among Microsoft's engineers, but the company now wants them to switch to GitHub Copilot CLI.

Microsoft is being investigated by the UK's CMA over its business software practices.

Old camera tech has been employed to form a 3D hybrid of NAND and DRAM tech using IGZO instead of silicon.

Under the new proposals, the lampposts in your street could double up as data centers with self-destructing, anti-tamper Nvidia chips.

Anthropic wants to help SMBs with purpose-built agentic workflows, skills and a US upskilling tour with free Claude Max trials.

Cisco CEO Chuck Robbins announces plans to lay off 4,000 workers as the company shifts its focus to high-growth areas.

New data reveals customer service workers spend more on trust, security and compliance than AI development itself.

Poor AI deployments are causing humans to spend a day a week as 'middleware' – liaising between siloed AI systems.

Amazon wants workers to be using more and more AI, so workers are just apparently wasting tokens to make it look like they are.

After using AI, many business leaders say are less likely to value their human workers, but the technology is still failing to deliver consistent ROI.

Privacy concerns aside, Meta workers are worried that AI-training mouse tracking software will ultimately lead to further job cuts.

Anthropic is adding new connectors and plugins to Claude to give it legal superpowers, and they're even available in Cowork.

Scientists just made the first ever observed organism with fewer than 20 amino acids in its make-up, and it was made possible by AI.

With minimal cable access, commercially available tools and AI, attackers can technically listen in to your conversations via your fiber optic cables.

OpenAI sets out plans to embed DeployCo workers inside customer enterprises to help them identify AI uses cases and transformation opportunities.

The more interactions an AI model has, the less reliable it becomes, experts find, as even the best only scored 80.9% – and the worst scoring just 10.0%.

New data shows that entrepreneurship is becoming more accessible – LinkedIn wants to help give SMBs a fighting chance.

US (and other global) hyperscalers could be pushed out of the running for contracts relating to sensitive public sector data.

Fedora and Ubuntu have set out plans to run small AI models locally, and this has caused a mixed reaction from the open source community.

Gemini will access more of your Gmail and Google Workspace context to product emails that sound more like you.

Three in four companies are meeting or exceeding AI ROI expectations, but security needs a bigger emphasis for future longevity.

Although most developers use AI daily, often or sometimes, some still use it rarely and many don't even use it at all.

AWS knows AI agents will be making payments for us in the future – and the first step starts with agent-to-agent micropayments.