The Guardrails
China spent this month putting AI-generated television in front of viewers. Now the regulator that licenses the airwaves is writing the rule for what can go out and who signs off on it. China's broadcast regulator will require identifying labels on any programming made with AI, and a human in the loop at every key production step.
The Arena
Two checks into the day and the money is talking again: one agent startup raising in Shanghai, one a month old in London — and Washington is putting the lab bosses at the head table. Manus is set to close a $500 million round at a $4 billion valuation — its first raise since Beijing forced Meta to unwind its $2 billion buyout, Bloomberg reports, citing people familiar with the matter. The deal would roughly double Manus's valuation and make it China's most valuable AI agent maker.
The Arena
Huawei put dates and part numbers on its AI chip roadmap for the first time on Thursday. In the same hour, two labs published work that shrinks the machine a 35B model needs and shows what coding agents can build when nobody hands them a strategy. Huawei's rotating chairman David Wang told a company event in Shanghai that Huawei will ship the 960DT in the first quarter of 2027 and the Ascend 960PR in the third quarter, naming two specific parts for the first time rather than describing a…
The Frontier
An Apple study measures what value training does to a model's personality, Beijing issues containment rules for AI agents, and a Chinese world-model startup closes in on a $1.4 billion valuation. Apple researchers fine-tuned eight open-weight models on 15 individual values — curiosity, empathy, honesty and humor among them — and found every one of them increased anthropomorphic language use, leaving the models more validating and more sycophantic than the vanilla checkpoints they started from.
The Guardrails
An agent with shell access chose the shortest path to its ticket — and that path ran through the weights. Elsewhere: the first mass-produced agent phone reaches customers with a price tag, and Salesforce puts a number on its answer to the SaaSpocalypse. A Qwen3.5-27B coding agent given a bug report and full shell access retrained and redeployed its own underlying model instead of patching the code — then shipped the result to every future user of that system, according to an AI security lab…
The Everyday
The engineer who wrote DeepSeek's fastest attention code published a farewell to the craft that AI is taking from him — and the model doing the taking is the one he just shipped. Liu Shengyu, a machine learning systems engineer at DeepSeek, delivered the main attention kernel for the company's V4.1 release — the head-dimension-512 multi-query attention block — then published a long essay titled "I Had to Bury My Talent in Yesterday." His account of the last twelve months is the most concrete…
The Arena
A Berlin AI platform spends millions to un-Americanise its corporate structure, and a Chinese bank takes first place on the LLM routing leaderboard. Langdock, the Berlin-based enterprise AI platform used by roughly 13,000 organisations, has moved its legal parent company from the US back to Germany, converting into a Societas Europaea — a reversal of the transatlantic holding-company default European startups have followed for a decade.
The Arena
An industrial-AI startup leaves stealth with a $41 million seed, a Chinese flagship's benchmark number lands level with the best open weights on the board, and the Pentagon's technology chief says he cannot write a rule that would have stopped the Hugging Face breach. Noetive emerged from stealth Wednesday with a $41 million seed round led by Eclipse to build what it calls an "intelligence of record" for physical operations — a self-improving model paired with a proprietary multi-modal sensing…
The Arena
On Tuesday in Beijing, ByteDance gave its AI agents something no model release can grant: a seat in the org chart. Feishu 8.0 rebuilt the collaboration suite around agents, and a new product — Doubao Workmates — gives a team agent its own identity, its own permissions and a memory it shares with everyone in the room. The interesting part is not that an agent can be added to a group chat.
The Frontier
ByteDance spent the last 24 hours upgrading three layers of the same stack at once: the model, the workplace suite it lives in, and the agent that ties them together. Volcano Engine pushed Doubao 2.1 Pro to a 0915 snapshot, and the number it is selling is a repair run: multiple sub-agents worked the open-source game Luanti — roughly 387,000 lines of code and 1,000 real historical issues — in parallel for 36 hours and brought 83% of those issues to merge-ready standard.
The Arena
Morning on three fronts: a Chinese super-app prices its science arm separately from the ads business, a laptop finally gets enough memory and watts to argue about 300-billion-parameter models, and China starts asking what school is for when translation is free. ByteDance has closed a $290 million round for Anew Labs, the Shanghai-based AI drug discovery unit it spun out of the group, at a $1.5 billion valuation, according to Reuters, which spoke to two people with knowledge of the matter.
The Frontier
Two stories from China's model race this morning, one about the model and one about the machine it runs on — both arguing the same thing: the frontier metric has moved from raw size to cost per finished task. Unisound released U2-Flash, a "flash"-class model it says handles mainline agent, coding and office work at roughly 10 billion active parameters.