China AI

Baidu apologises in China's first AI defamation case — two years on

The Guardrails

Baidu apologises in China's first AI defamation case — two years on

A lawyer who searched his own name and found a criminal record that never existed finally got a written apology today. The reasoning behind the order is the part the industry should read. Baidu formally apologised in writing to Nanjing lawyer Li Xiaoliang on September 22, at the Nanjing Jiangbei New Area People's Court — the first time a Chinese court has forced a model operator to apologise for what its model said about a named person.

Deep Dive — Alibaba wants to own the whole stack. Memory says no.

The Arena

Deep Dive — Alibaba wants to own the whole stack. Memory says no.

The most important number Alibaba put on stage in Hangzhou this week was not ten trillion. Alibaba opened its annual Apsara Conference on Tuesday with a plan that is unusually easy to restate, because the company said it in one sentence: it will train Qwen models of five trillion to ten trillion parameters, run them on its own accelerators wired into clusters of up to 500,000 cards, and operate more than 20 gigawatts of data centre capacity worldwide by 2032.

Kimi Delta Attention gains 2D rotations from a signed gate

The Frontier

Kimi Delta Attention gains 2D rotations from a signed gate

A two-number change to one of the open field's most widely used attention modules buys a provable class of expressivity at no extra cost — and the authors are unusually clear about what it does not buy. Allowing Kimi Delta Attention's channel-wise gate to go negative — into [−1, 1] — and its delta-rule coefficient up to 2 lets the module realise 2D rotations, giving it the state-tracking expressivity of DeltaProduct-2 while its transitions stay diagonal-plus-rank-one and non-expansive.

Alibaba plans a 10T-parameter model and a 500,000-chip cluster

The Arena

Alibaba plans a 10T-parameter model and a 500,000-chip cluster

Alibaba's cloud conference in Hangzhou opened this morning with a three-part commitment that is harder to wave off than the usual keynote arithmetic: a model bigger than anything trained so far, the chip to run it, and the power to feed both. Huawei spent the same week putting a coding model inside its own operating system, and a small project on the local-AI forums quietly showed that a model's refusals can be switched off without touching a single weight. Alibaba says its next Qwen models…

22 leaders signed a frontier-AI control pact. The US, China and UK didn't.

The Guardrails

22 leaders signed a frontier-AI control pact. The US, China and UK didn't.

Two things landed on the same Monday: a coalition of middle powers published the sharpest international statement yet on frontier AI control, and a Lean-verified number-theory proof appeared with one anonymous account standing behind it. Finland's president, Alexander Stubb, published a joint statement on Monday — a day before the UN General Assembly's high-level debate opened in New York — demanding that frontier AI stay under human control, and it comes with three concrete asks.

Today in AI — September 21, 2026

The Guardrails

Today in AI — September 21, 2026

Monday's through-line was accountability: who answers when a model misbehaves, and who pays when it works. Washington drew that line at management, two labs reportedly negotiated to attack each other's models, China's flagships kept winning the volume race while one of them took a shelf at Amazon, and two compression papers argued that smaller can now mean better rather than merely cheaper. Models & Research * ByteDance Seed and Tsinghua AIR open-sourced DAPO, a reinforcement-learning…

OceanBase tops Berkeley's data-agent benchmark with GLM-5.2

The Frontier

OceanBase tops Berkeley's data-agent benchmark with GLM-5.2

Two benchmark results and one hype check to start the week: a Chinese database outfit took the top of a Berkeley-run data-agent leaderboard on a Chinese model, a frontier model reportedly broke a century-old naval cipher, and the robot industry's own trade body published the least flattering number of the year. OceanBase, the distributed database business Ant Group spun out, has taken first place on the Data Agent Benchmark with 90.62% accuracy — the first submission to clear 90%, and it was…