Alibaba open-sources Qwen-UI-Agent trained on 100 real phones

Two stories caught our eye this morning: Alibaba's bet on putting AI inside real devices, and a reminder that AI policy is now a campaign war chest.

Alibaba open-sourced Qwen-UI-Agent, a GUI agent model trained and evaluated inside a fleet of more than 100 physical phones rather than clean simulators. The model is built to drive devices the way a person would — tapping, typing, running command-line actions, and calling APIs to finish multi-step tasks across mobile, desktop, and web. Alibaba says it tops five of six major GUI benchmarks, beating GPT-5.6 Sol, Claude Opus 4.8, and ByteDance's Seed 2.1 Pro, and that its 27-billion-parameter main version reaches 92.2% success on a real-device test set and 97.5% on everyday Android tasks.

The detail that matters is the training environment. Most GUI agents score well in sandboxes but fall apart on real phones — pop-up ads, expired logins, shifting layouts. Alibaba attacked that by standing up a runtime of 100-plus real phones running 150-plus apps, with a health monitor that auto-switches devices on failure and virtual-screen tech that multiplies concurrency about 20x, then ran a three-stage reinforcement-learning loop where an AI data flywheel writes tasks, verifies results, and diagnoses failures with humans only supervising.

Our take: this is the sim-to-real gap being closed the expensive way. A 27-billion-parameter model out-executing frontier flagships on real devices suggests the bottleneck for useful agents isn't raw scale — it's training on the messy truth instead of tidy mocks. We've tracked Alibaba's widening AI footprint before, including Apple's China AI work built with Alibaba's support.

AI, crypto, and online betting companies have become the dominant spending forces in the 2026 US midterm races, with corporate outlays hitting a record $517 million. Reuters reports the three industries now lead a reshape of campaign finance, as crypto firms in particular pour money into races that will decide who writes the next round of AI and digital-asset rules. Independent coverage from International Business Times Singapore frames the same shift as a "new corporate kingmaker" effect, with crypto's election playbook spreading into AI and gambling.

Why it matters: AI is no longer just the subject of policy — it's a donor with a seat at the table. The companies most affected by upcoming AI legislation are now among the biggest spenders shaping who gets to write it, which raises the stakes on every regulatory fight from model testing mandates to frontier safety rules.

Should AI and crypto firms be allowed to spend their way into shaping the laws that govern them? Tell us in the comments.

Sources: Qwen-UI-Agent (GitHub)

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