California opens highways to self-driving truck testing

Share
California opens highways to self-driving truck testing

Two very different AI frontiers moved today: California finally let autonomous trucks onto its public highways, and a ChatGPT core contributor pulled in one of the biggest Pre-A rounds in Chinese embodied AI.

California's DMV has granted testing permits to Aurora Innovation and Kodiak AI, clearing autonomous heavy trucks to operate on public roads in the state for the first time — and Kodiak has already started, running a handful of test trucks around its Mountain View office. The permits, issued under rules the DMV approved April 28 that lifted California's long-standing ban on driverless vehicles over 10,000 pounds, require a human safety operator behind the wheel and bar testing on roads with speed limits of 25 mph or less unless a truck is on a "direct route" between destinations. Both companies are California-born but had been forced to build their commercial operations in Texas under their home state's ban: Aurora launched a Dallas–Houston self-driving truck service in May 2025 and has since added routes across the Southwest, while Kodiak started commercial driverless operations in West Texas in January 2025 and expanded to on-highway runs. The homecoming is not uncontested — last week, Teamsters California sued the DMV in Alameda County Superior Court, arguing the agency skipped legally required study and public disclosure of the economic fallout and ignored safety risks to motorists, with the union warning the technology threatens hundreds of thousands of driving jobs.

The stakes are real on both sides. California is the last big holdout for autonomous trucking in the US, so this is the final domino in the regulatory map — but it also hands the industry its toughest political fight yet, in a state where the Teamsters have genuine leverage. The safety-operator requirement keeps this firmly in the testing phase, and the lawsuit means the path from test permits to commercial deployment will be litigated in public, not just engineered. We've been tracking the wave — Pony.ai's fourth-gen driverless heavy truck enters mass production — and California joining the map is the milestone that matters most for US trucking.


Liangyuan Xinchuang, an embodied-AI startup founded by Jiang Xu — described by Chinese media as a core contributor to ChatGPT and a former OpenAI algorithm expert — has closed a Pre-A round of several hundred million yuan, an unusually large sum for a round this early. Sina Finance reported in an exclusive that state-linked Guoke Investment, the venture arm affiliated with the Chinese Academy of Sciences, led the round, with Phoenix News reporting that state-owned conglomerate China Merchants Group is also among the backers. The company is building Physical AI foundation models for embodied intelligence, joining a crowded Chinese field racing to make robots that can perceive and act in the physical world. For a Pre-A, the check size signals that Chinese state capital is treating embodied AI foundation models as a strategic bet — and that the talent pipeline from OpenAI to Chinese robotics startups is still flowing.

What to watch: Aurora's first California test runs, and the next hearing in the Teamsters' suit against the DMV.

The Teamsters call driverless trucks an existential threat to hundreds of thousands of driving jobs; the industry says the safety data is already in. Whose argument wins in California? Tell us in the comments.

Read more

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

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

Suleyman says Microsoft's base model beat OpenAI's on a quarter compute

Suleyman says Microsoft's base model beat OpenAI's on a quarter compute

Microsoft finished pre-training its biggest in-house model yet — then used it to make a point about the company it bankrolls. Microsoft AI completed pre-training of MAI-Base-2, a mixture-of-experts foundation model with 80 billion active and 2.8 trillion total parameters, and says the recipe is 4.2–12.3x more pre-training FLOPs-efficient than DeepSeek-V4-Pro-Base. The run itself is the number worth keeping: 16,000 Nvidia GB200 GPUs for 38 days across August and September, at 21% average model

Amazon's new Bee devices listen all day — for $11.99 a month

Amazon's new Bee devices listen all day — for $11.99 a month

Amazon turned its always-listening Bee line into a two-device family on Thursday, and NVIDIA opened a 2.8-trillion-token Indic corpus recipe. The connecting thread: both are bets that the next AI money is in memory — yours, or your model's. Amazon put two new always-listening devices on sale, and both now lead to a subscription. Bee Frontier, the next-generation wristband that also clips to clothing, ships now at $74.99 in Onyx, Lavender and Sage — up from the $50 predecessor — with seven-day

Perplexity open-sources its agent sandbox after 118 failed escapes

Perplexity open-sources its agent sandbox after 118 failed escapes

Perplexity handed its agent containment layer to the open-source world, and Washington's AI task force stopped reading briefings and started working. Perplexity open-sourced SPACE on Thursday — the sandbox it built to keep its own AI agents from getting out — and says nine frontier models failed to break out of it across 118 attempts. Each agent runs inside a small virtual machine with its own operating system, with credentials and network access managed outside the sandbox so an agent can re