Artificial intelligence, cyber security, digital policy and innovation.
US prosecutors have charged a former OpenAI engineer with allegedly installing a recording device in an executive conference room. The case highlights a security problem that model safeguards cannot solve: trusted employees can reach the conversations and systems surrounding frontier research.
14 October 2026 · 3 min readASML says customers want more chipmaking equipment than it can supply through 2028, with some discussing orders for 2030. The constraint shows that AI expansion depends on a narrow and slow-moving layer of industrial infrastructure.
14 October 2026 · 3 min readNvidia will provide UK researchers with $250 million of computing access over five years for agentic AI safety and security work. The pledge could accelerate testing, but it also highlights how public-interest research depends on commercial infrastructure.
14 October 2026 · 3 min readNvidia’s share of Chinese data-centre AI accelerators fell to 34% as restricted supply and domestic alternatives shifted competition from individual chips toward complete computing systems.
13 October 2026 · 3 min readA proposal backed by about 400 researchers would introduce common containment standards, independent supervision and a registry of serious AI security incidents.
13 October 2026 · 3 min readAnthropic’s tests suggest small models are approaching the ability to copy themselves across poorly secured connected devices, challenging security built around centralised services and shutdown controls.
13 October 2026 · 3 min readScaling its own inference chips to multi-gigawatt deployments could reduce Microsoft’s dependence on external accelerators and give the cloud group greater control over cost, energy and product design.
12 October 2026 · 4 min readWarnings from Singapore’s president and new IMF labour-market work point to a structural risk: entry-level jobs may shrink before education and training systems adapt.
12 October 2026 · 4 min readCompulsory reporting would turn frontier-AI safety from a voluntary promise into an operational duty, with major consequences for testing, disclosure and accountability.
12 October 2026 · 3 min readThe next phase of AI growth may be constrained less by access to memory chips than by the money, power and physical infrastructure required to deploy computing at scale.
12 October 2026 · 2 min read