Travis Kalanick returns to robotaxis, Apple's Siri AI fades fast, and AI lawsuits keep stacking up

Picture this: the man who built Uber, got pushed out, and spent years in the wilderness - and now he might be circling back to finish what he started with self-driving cars. Meanwhile, the publisher lawsuit queue just got two names longer, and Apple's big Siri comeback quietly flatlined in someone's pocket. A lot to unpack today.

Travis Kalanick's new company Atoms isn't his first post-Uber venture. What was the name of his cloud kitchen startup that he founded after leaving Uber in 2017?

The answer is hiding near the bottom of today's newsletter... keep scrolling. 👇

Travis Kalanick, the co-founder who built Uber into a global giant before being forced out in 2017, may be lining up his next act in the very market he helped shape. His startup Atoms - which he has described as a way to complete "unfinished business" - is reportedly exploring entry into the robotaxi space, according to TechCrunch.

Kalanick hasn't publicly confirmed specifics about Atoms' robotaxi plans, but the framing is hard to miss. Uber's self-driving ambitions eventually became Waymo's market to own, and London just got its first robotaxi service via Wayve on Uber's platform. The timing of Atoms' interest is striking - the window for new entrants may be narrowing fast as incumbents lock in partnerships.

The robotaxi market is brutally capital-intensive. Waymo, backed by Alphabet's deep pockets, has a multi-year head start in San Francisco and Phoenix. But Kalanick built a global ride-hailing network once before by moving faster than incumbents expected. Whether Atoms has the technology stack to compete - or plans to partner with an existing AV developer - remains the big open question.

Read the full story →

Atoms is still early-stage and unconfirmed, but if Kalanick does enter robotaxis, expect him to move aggressively - that's always been his playbook.

Wired spent the summer with the beta version of Apple's revamped Siri AI - and the verdict is quietly brutal: by the time the full release approached, the reviewer had forgotten Siri AI even existed. The piece, published this weekend, is less a takedown and more a portrait of a product that generated genuine early excitement and then... faded into the background.

The initial beta impressed enough to spark interest - Apple's pitch was a more capable, context-aware Siri that could actually do things across your apps. But somewhere between beta and the approaching full release, the magic wore off. The reviewer drifted back to other tools and stopped reaching for Siri at all.

This is a problem Apple has struggled with for years: Siri is rarely the first tool people reach for, and competing assistants have raised the bar significantly. For Siri AI to succeed, it needs to become habitual - the kind of tool you use without thinking. Right now, at least for one extended tester, it's still the kind you forget is there.

If you're curious about how AI assistants compare in practice, check out our AI Tools coverage for more breakdowns.

Read the full story →

Apple's real challenge with Siri AI isn't technical - it's behavioral. A capable AI that nobody remembers to use is just as useless as one that can't do anything.

The Seattle Times and Newsday are the latest news organizations to sue OpenAI and Microsoft over the use of their journalism to train AI models. The lawsuits, reported by both The Verge and TechCrunch, follow a now-familiar pattern: news organizations argue their copyrighted work was scraped without permission or compensation to build AI systems that now compete directly with them for readers and advertising revenue.

This wave of AI copyright litigation is accelerating. The New York Times filed the landmark case against OpenAI and Microsoft back in late 2023, and dozens of publishers have followed. Each new filing adds legal pressure and potentially discovery risk for the AI companies involved - revealing more about how their training pipelines actually worked.

Separately, there's a related dispute brewing around Anthropic's settlement with authors. TechCrunch reports that authors are pushing back because publishers and agents appear to be claiming more than their fair share of the settlement payments - meaning the writers whose actual work was used may end up with less than expected. The legal battles around AI training data are getting messier, not cleaner.

Read the Seattle Times/Newsday story →

Read the Anthropic settlement story →

These lawsuits are still years from resolution, but every new filing makes it harder for AI companies to argue that scraping journalism was a legally settled question.

Here's a use case nobody asked for: AI-generated food photography on restaurant menus. The Guardian reports that consumers are increasingly encountering AI-generated images that make food look genuinely revolting - meats described as resembling leather belts, bread with reptile-like texture, and colors that register as deeply wrong to human eyes. One person was so alarmed she shared the images with her 26,000 followers on X.

The problem is the uncanny valley applied to food. AI image generators are trained to produce visually impressive outputs, but they often get textures, reflections, and proportions subtly wrong in ways that trigger immediate disgust rather than appetite. For restaurants, which adopted AI images presumably to save money on food photography, the result can be the opposite of the intended effect.

This is a useful reminder that AI image generation has clear limits when accuracy and trust matter. A professional food photo has to make someone want to eat the dish. An AI image that looks almost right - but not quite - does the opposite. The restaurants saving on photography costs may be paying for it in lost orders.

Speaking of building things quickly with AI - if you're working on a website or landing page, 60sec.site lets you spin up an AI-built site in under a minute. Worth a look if you want something that actually looks good.

Read the full story →

AI image tools need a human review step before they go anywhere near customer-facing materials - the gap between "looks impressive" and "looks appetizing" is bigger than most restaurant owners realize.

Industry leaders are pushing back on fears of job automation in the design sector, with The Guardian reporting that business leaders are framing AI as "the intern in the office" rather than a replacement for skilled staff. Design and film production companies, as well as manufacturers, are all adopting generative AI at pace - but industry bodies say professional designers shouldn't feel threatened.

The argument is familiar: AI handles the repetitive, low-value work, freeing up human designers for the creative and strategic decisions that actually require taste and judgment. It's a comforting framing, and there's real truth to it for experienced professionals. The harder question - what happens to entry-level designers who used to build those foundational skills doing exactly the work AI now handles - gets less airtime.

Read the full story →

Senior designers are probably fine. The worry is for people who haven't yet become senior designers - and AI is quietly closing the door on the traditional path to getting there.

The answer is CloudKitchens! After being pushed out of Uber in 2017, Travis Kalanick founded CloudKitchens, a company that converts commercial real estate into shared kitchen spaces for delivery-only food brands. It raised over $400 million and expanded globally before Kalanick shifted focus to Atoms.

Today's food photo story got me thinking: have you come across an AI-generated image recently that was so obviously wrong it put you off a product or service? Hit reply and tell me what it was - I read every response and I'm collecting the best examples.

That's it for today. For more daily AI coverage, head to dailyinference.com - and we'll see you tomorrow with more. 👋