AI meets the courtroom and the border checkpoint: This week's fight over the rules of the game

The question of who gets to decide the rules for AI won't be settled in one place. This week, the attempts came on many fronts at once: in court, in government buildings and in the draft legislation itself.

The lawsuits keep piling up

Most of the headlines came from the courtrooms. The Elvis Presley estate and other rights holders have sued Anthropic, alleging that the Claude AI model was trained on Presley's music without permission. It is a copyright question: can you use other people's works as training material without asking?

At the same time, OpenAI is under pressure from several directions. Texas has sued the company and CEO Sam Altman for misleading the public about ChatGPT's safety, and is demanding, among other things, that ChatGPT not be offered to minors in the state. Arkansas has filed a similar suit, which according to sources is one of several from US states. And a parent from Hawaii claims that children's information was used for training without consent. OpenAI denies this, and nothing has been tested in court. It's worth remembering: these are still allegations.

Governments pick sides

On the other side of the table, states are acting as customers. From October 26, the Swedish Government Offices are blocking ChatGPT and Claude on work computers for reasons of information security, pointing employees to Microsoft's Copilot instead. It is a somewhat different approach from making laws: you simply decide what employees are allowed to open.

The same logic can be seen in the US, where the Department of Homeland Security (DHS) has banned China's DeepSeek in civilian federal agencies. The stated reason is the risk of data leaks and malicious code. The ban also covers products built on top of DeepSeek's models, but it does not affect private companies or ordinary users. The order sets no deadline and leaves it unclear how far it reaches. A few days later, a new DeepSeek model was released that, according to the company itself, can match Claude Opus 5 at a fraction of the price. China is going the opposite way and has proposed requiring advance approval of advanced AI models before they are shared abroad.

Who sets the rules?

One interesting development is that even the AI companies are moving into lawmaking. According to Politico, Anthropic has shown lawmakers a draft that would give the government the power to halt the release of models posing an imminent risk of serious harm. It can be read as responsibility, but also as an attempt to shape the rules before others do.

Trust in the tools is also being tested on a smaller scale. In the UK, a detective is being investigated for allegedly using Microsoft Copilot inappropriately in criminal cases, but no wrongdoing has been established.

What can you do?

For ordinary users, the takeaway is that AI tools can be banned, changed or restricted on short notice, especially in the workplace. Check what your employer allows before you paste confidential information into a chatbot. Also keep an eye on how the US lawsuits end, and on whether regulators follow the example of Sweden and the US. That will largely determine which tools the rest of us get to use.