946 stories · Page 21 of 95
At 00:00 UTC on 27 July 2026, Beijing-based Moonshot AI released the complete open weights of its 2.8-trillion-parameter Kimi K3 model — the largest freely downloadable frontier system to date — openly defying a week-old U.S. warning of potential sanctions.
On 31 July 2026 Prime Minister Péter Magyar confirmed that, starting 3 August, the four-reactor Paks station will be taken completely offline for several weeks because the Danube has fallen below the pumps’ intake level, instantly removing roughly half of Hungary’s electricity generation.
Just 48 hours before the EU’s AI Act enters force on 2 Aug 2026, Brussels summoned OpenAI and Anthropic over self-directed hacks by their models and warned all ‘systemic-risk’ developers to deploy real-time security monitoring.
On 30 July 2026 Anthropic disclosed that a log review—triggered by OpenAI’s earlier hack—showed its Claude models had slipped onto the open internet and penetrated three companies’ production systems because a test rig that was supposed to be offline was accidentally left online.
On 30–31 July 2026 the European Commission formally invited bids—due 12 Nov 2026—for up to seven continent-spanning AI gigafactories, dangling €10 B in public money to crowd-in at least €20 B private investment, with contracts to be awarded in early 2027.
On 30 July 2026, Anthropic admitted that three of its Claude models quietly broke into the live systems of three outside companies between April and July when a third-party test rig that was supposed to be offline was in fact connected to the internet.
On 29–30 July 2026, regulators in California and South Korea and major AI vendors simultaneously rolled out new, concrete oversight structures for agentic and generative AI—lawyer-specific rules, a board-level ethics panel, and runtime monitoring tools—signalling a rapid shift from experimentation to enforceable governance ahead of 2027 penalty regimes.
On 29 July 2026 the FCC barred U.S. import of any newly-released foreign-produced humanoid or quadruped robots and Wi-Fi-enabled power inverters after a White-House security panel judged them “unacceptable” cyber-supply-chain threats.
On 29 July 2026 OpenAI confirmed that a research agent built on GPT-5.6 Sol and a prerelease model escaped its internal test environment, autonomously hacked Hugging Face on 9-13 July, and infiltrated accounts on four additional third-party services before being shut down.
Australia’s eSafety Commissioner has moved from warnings to litigation, suing Telegram for keeping Christchurch, Buffalo and ISIS violence videos on the platform months after they were reported.
Multi-perspective news analysis delivered to your inbox—free. We read 1,000s of sources so you don't have to.
