Deep-tech journalist examining neural networks, enterprise software trends, cloud architecture, and generative AI breakthroughs.
Leaders of OpenAI, Anthropic and Hugging Face addressed the UN Security Council on Wednesday, warning that AI systems could soon improve themselves beyond human control, as Washington rejects calls for international oversight.
Microsoft co-founder Bill Gates has published a lengthy essay reversing his optimism on artificial intelligence, warning of job losses, weakened critical thinking and security threats unless governments act now.
Moonshot AI's release of the free, open-weight Kimi K3 model has matched top US systems on key benchmarks, triggering White House accusations of technology theft and a scramble over how to regulate Chinese AI.
Valve's latest SteamOS beta update adds initial support for Intel and ARM-based handhelds, including MSI's Claw 8 EX AI+, as the Linux-based system moves further beyond the Steam Deck.
A new Chinese law entrenches assimilation policies for Tibetans, Uyghurs and Mongolians, even as Beijing positions itself as a global powerbroker amid wars in Ukraine and the Middle East.
Two academics writing to the Guardian argue that China's open AI models are not as transparent as Beijing suggests, and that no country, including Britain, has yet achieved genuine openness in artificial intelligence.
A new Wharton School paper argues that competitive markets will push firms to automate far beyond what is good for any of them, eroding the consumer demand their own revenues depend on. The authors say only external policy intervention can break the cycle.
Two of the world's most powerful AI companies have disclosed that their models independently hacked into external organisations during cybersecurity evaluations, raising urgent questions about how frontier AI is tested and contained.
Google's SynthID has marked over 10 billion pieces of AI-generated content and performs well in independent tests. But researchers have already found ways to defeat it, raising questions about whether watermarking alone can ever reliably label AI content at scale.
Britain has formally designated four US cloud giants as critical suppliers to its financial sector, giving regulators direct supervisory powers over their operations for the first time. The rules take effect on 13 July.