ByteDance creates AI data and safety division

ByteDance is reorganizing around the data problem, Europe's biggest airline is betting its operations on Google's models, and Chinese robotaxis are expanding deeper into the Gulf — a morning of big industry moves.

ByteDance has reportedly created a new first-level "AI data and safety" division — its first new AI unit since Seed and Flow were founded in late 2023. Chinese tech outlet IT之家 reported Tuesday, citing multiple sources, that the unit sits parallel to Seed, Flow and Douyin, with former TikTok platform-responsibility and live-streaming chief Adam Wang in charge — a leader whose live business was one of TikTok's biggest revenue engines. The division grows out of Global Data, the roughly 100-person team created in 2023 to source and quality-check training data for Seed's models, and it will now serve all of ByteDance's large models across the full data pipeline: standard-setting, procurement, synthetic cleaning and quality evaluation.

Putting data quality and safety on the same organizational level as the model teams is a signal that ByteDance now treats the data pipeline as a competitive lever, not a support function — and it arrives as the company reportedly pre-trains a 10-trillion-parameter model to rival Anthropic (ByteDance pre-trains a 10T-parameter model to rival Anthropic). ByteDance hasn't commented publicly, so the org chart should be read as reported, not confirmed.

Ryanair, Europe's largest airline, signed a five-year data and AI partnership with Google Cloud that puts Gemini tools and DeepMind models to work on crew scheduling and operational decisions. The deal announced Wednesday brings Google Workspace and Google Cloud services to 35,000 employees as the carrier grows toward 300 million passengers a year, and it complements Ryanair's existing use of AWS — a deliberate multi-cloud hedge against outages. Financial terms were undisclosed. Airlines are the kind of customer where generative AI has to prove itself in operations: a scheduling error is measured in cancelled flights, not tokens.

Chinese robotaxi operators are deepening their push into the UAE, where Baidu's Apollo Go runs commercial service across about 25 square kilometers of Dubai's Jumeirah district. Apollo Go's UAE head told Xinhua the operator plans to expand to hundreds of square kilometers and a fleet of hundreds of vehicles by year-end; WeRide and Pony.ai are also operating in the Emirates. Dubai wants 25% of its transport autonomous by 2030, and Chinese companies are the early movers chasing that target — helped by Gulf regulators who move faster than most Western markets.

What to watch: Whether ByteDance confirms the new division's charter — and whether Ryanair's Gemini bet pulls other carriers off the cloud sidelines.

Does data quality — not model architecture — decide the next round of frontier AI? Tell us in the comments.

AI 101 — How do AI video generators work?

An AI video generator is a model that starts with a grid of random noise and repeatedly refines it into a moving picture that matches your prompt — the same trick image generators use, stretched across time. Nothing is filmed, and nothing is stitched together from clips the model has seen. The pixels in the finished video were invented, frame by frame, all at once. Why it matters right now Kling 4.0, Kuaishou's flagship video model, went fully live on October 15 with up to 30 seconds of vid

Houmo moves compute inside memory — and A20 enters production

Beijing chipmaker Houmo used its partner conference this week to lay out two compute-in-memory routes at once — and to say the first chip built on the newer one is already in production. Houmo's A20, its first chip built on a 3D compute-in-memory architecture, has entered production. The part supports up to 16 chips fully interconnected, claims a 50% improvement in compute utilization over the previous generation, and covers models from 3B to 30B parameters, according to co-founder and product

Kling 4.0 goes live with 30-second native video at 10-bit HDR

Kuaishou's Kling AI has spent two years becoming the video model people actually finish projects with. Its new flagship, Kling 4.0, went fully live on October 15 — and this round is less about prettier frames than about the two things that make a generated shot usable in a real edit: length and control. Kling 4.0 generates up to 30 seconds in a single native pass, outputs 10-bit HDR at 4K, 1080p and 21:9, and takes up to 15 reference assets in one task. That budget splits across ten images, fi

Brain organoids classified Braille at 83% — in a trio, not alone

Two stories today: human brain tissue on a chip decoded a tactile signal, and the data-centre company behind some of the largest AI contracts in Europe turns out to be six months behind on its flagship site. Human forebrain organoids grown on a microelectrode array classified Braille letters with 61 percent accuracy on their own and 83 percent when the responses of three organoids were combined, in a peer-reviewed study from the University of Bristol. The team took event-based tactile readings

Read more

AI 101 — How do AI video generators work?

Houmo moves compute inside memory — and A20 enters production

Kling 4.0 goes live with 30-second native video at 10-bit HDR

Brain organoids classified Braille at 83% — in a trio, not alone