OpenAI, Anthropic, Google and SpaceXAI sued for allegedly agreeing to slow down AI: what it means for businesses in Vigo and Galicia

On 18 September 2026, four paying subscribers to ChatGPT, Claude, Grok and Gemini filed a class-action antitrust lawsuit against OpenAI, Anthropic, Google and SpaceXAI in the federal court for the Northern District of California. They accuse the companies of agreeing to slow down the development of artificial intelligence, which they say reduces what they get in return for their subscription. For a business in Vigo or Galicia, nothing changes immediately, but it's a good moment to stop relying on a single AI tool.

It's rare to see something like this: companies being sued for wanting to slow down. A year ago the argument ran the other way, whether AI was moving too fast. Here are the facts as reported, what each side says, and what it means if you use AI every day at work.

What exactly happened?

The timeline, as reported in the lawsuit and by outlets such as OPB, Bloomberg Law and MediaNama, runs as follows:

According to MediaNama, the lawsuit asks the court to certify it as a nationwide class action, seeks treble damages as provided for under US antitrust law, and requests an injunction blocking the pact. At the time the news broke, the companies had not yet responded.

What are the claimants alleging?

That an agreement between competitors to limit what they produce is illegal, even if done with good intentions. They rely on Section 1 of the Sherman Act, the US antitrust law that bans rivals from colluding to restrict competition.

What's interesting is what they're not asking for. They don't object to each company slowing down on its own, nor to Congress passing a law requiring a slowdown. What they reject is the shortcut: the decision coming from a private agreement between the four most powerful companies in the sector, rather than from each company's own choice or from legislation.

Lead counsel Nick Rowley summed it up with a striking idea: he accepts that AI can be dangerous, and precisely because of that, he argues its safety can't be left to private agreements between profit-making companies.

Does the lawsuit have legs?

There are two readings, and both have a point.

• It's opportunistic. Consumer harm is hard to prove: no one has shown that ChatGPT or Claude are worth less today than before Amodei's essay. And what the claimants are effectively asking for, continued acceleration, is exactly what worries many researchers.
• It has legal grounds. In competition law, the motive behind a pact usually doesn't matter. If competitors agree to limit their output, the agreement is suspect, even if it's done for everyone's safety.

The result is a pincer for the labs. If one slows down alone, its rivals overtake it. If they slow down together, they get sued. The only clean way out is for a state to decide. That's why some believe the lawsuit, far from blocking AI regulation, makes it almost inevitable.

What about Europe? There's already a law here

In the European Union, the underlying question, who decides how much AI slows down, has a different answer: the law decides. The EU AI Act (Regulation (EU) 2024/1689) came into force in August 2024 and applies in phases. It bans certain uses, requires transparency from general-purpose models, and sets obligations for high-risk systems.

And there's a detail close to home: AESIA, the Spanish Agency for the Supervision of Artificial Intelligence, the body that oversees compliance with that law in Spain, is based in A Coruña. Galicia is home to the whole country's AI watchdog.

What does this mean for a business in Vigo or Galicia?

In the short term, nothing. Your ChatGPT, Claude or Gemini subscription will keep working exactly the same tomorrow. But the case leaves three practical lessons:

1. Don't rely on a single provider. The big AI companies can change prices, limits or features from one month to the next, whether by their own decision, a pact or a court ruling. If your business uses AI to handle customers, draft quotes or summarise emails, make sure you can switch models without having to rebuild everything.
2. Write down what you do with AI. Which tools you use, with what data, and for what purpose. If a client, an inspection or AESIA itself ever asks, you'll have it to hand. And if you process personal data, GDPR still applies.
3. What really sets you apart is still yours. A clear website, well positioned and easy for AI to read and cite, doesn't depend on what four companies in California decide. I cover this in how to appear in ChatGPT and AI answers.

If you want to bring AI into your business without being tied to a single brand, for example with an assistant that answers your customers or an automation that saves you paperwork, I can help from Vigo. Take a look at how I build custom apps for businesses or request a quote.