AI’s Big Players Are Building a Single “Do Not Scrape” Switch for Publishers
Today’s AI news has a common thread running through it: trust. Who do you trust with your content? With your money? With claims about what a technology can actually do? These questions are shaping how AI develops — and today’s stories offer a useful map of where the fault lines are forming.
Major AI Companies Are Designing a Shared Opt-Out System for Web Scraping
Web scraping — the automated process of copying content from websites to train AI models — has been a source of tension between publishers and AI companies for years. Now, The Verge reports that OpenAI, Google, and other major AI players are working together on a shared system that would let publishers opt out of scraping across all participating companies at once.
Right now, if a publisher wants to stop multiple AI companies from using their content, they have to contact each one separately — filing different requests, navigating different processes, often getting different results. The proposed system would work more like a single switch. Flip it, and every company participating in the agreement agrees to honor it.
For bloggers, news organizations, photographers, and anyone else who publishes online, this could matter quite a bit. Instead of playing whack-a-mole with individual AI labs, you’d have one place to declare your preferences. Whether the system actually works in practice — and which companies sign on — will determine how meaningful it ends up being.
Why this matters: A shared opt-out standard would shift power slightly back toward creators. But it only works if the biggest scrapers actually follow it.
“Shared system lets publishers opt out of scraping across multiple AI companies at once.”
A New App Wants AI to Handle the Group Chat Restaurant Debate
Anyone who has spent 45 minutes in a group text trying to agree on dinner knows the problem Zest is trying to solve. According to TechCrunch, the startup has built an AI assistant that collects everyone’s food preferences via text, helps the group vote on options, and then contacts the restaurant directly to book a table — no phone call required.
This is what’s often called an “AI agent” — a system that doesn’t just answer questions but actually takes actions on your behalf in the real world. Zest’s agent reaches out to restaurants the same way a person would, completing the whole task rather than handing it back to you halfway through.
For people managing group dinners, work lunches, or family outings, the appeal is real. The friction of coordinating preferences and then actually securing a reservation is exactly the kind of repetitive task AI is well-suited to handle. The bigger question is reliability: restaurants change availability quickly, and an AI making reservations autonomously would need to handle rejections, alternatives, and confirmations without dropping the ball.
Why this matters: This is a practical, everyday example of AI agents moving from demos into tools people might actually use on a Friday night.
“AI assistant considers everyone’s preferences and automatically contacts restaurants to book tables.”
A Theranos Whistleblower Sees a Familiar Pattern in AI Hype
Tyler Shultz, the man who helped expose the fraud at Theranos — the blood-testing startup that claimed it could run hundreds of medical tests from a single drop of blood — is now looking at the AI industry and feeling a sense of déjà vu. TechCrunch spoke with Shultz, who argues that today’s AI boom creates conditions that are ripe for similar deception.
His concern is structural. When companies with no profits carry valuations worth billions of dollars, there’s enormous pressure to justify those numbers. Theranos collapsed because it promised things its technology couldn’t deliver. Shultz warns that the same incentive — keep investors happy, keep the story going — could push some AI companies to exaggerate what their products can actually do.
For anyone investing in AI companies, using AI tools in their business, or making decisions based on AI vendor claims, this is a useful lens. It doesn’t mean every AI company is fraudulent. It does mean the pressure to oversell is very real, and scrutinizing actual results matters more than listening to pitch decks.
Why this matters: Hype cycles don’t always end in fraud, but they often end in disappointment. Knowing what questions to ask protects you either way.
“AI boom creates similar risks because companies with no profits are being valued at enormous sums based on hype.”
Also Happening in AI
OpenAI quietly pushed out version 3.42.0 of its Python library on October 31st, adding new features for developers building AI agents. Separately, Nvidia is reportedly weighing whether to help finance an AI data center being built by Nscale in West Virginia, according to TechCrunch — a sign of how deeply chip companies are now entangled with the infrastructure side of the AI business. OpenAI CEO Sam Altman also made waves by suggesting, per The Verge, that it makes more sense to keep computing power concentrated in data centers than on personal devices. Meanwhile, Ars Technica reports that despite the buzz around AI agents, almost nobody is actually using them to order DoorDash — a useful reality check on adoption. And Starbucks is testing AI-powered drive-thru menus that recognize returning customers and show personalized options, according to The Verge.
What to Watch
The opt-out scraping system is one to follow closely over the next few months — specifically, which companies sign on and whether independent publishers report it actually working. Shultz’s warnings deserve attention in parallel: watch for any AI companies facing regulatory scrutiny over inflated capability claims, particularly those with large valuations and thin revenue. The gap between what AI is marketed to do and what it reliably does in the real world is where the most consequential stories of the next year are likely to emerge.