Everyone talks about slowing AI down, but the rules fail to materialize

There is something paradoxical about this week's stories on AI and politics. Both tech leaders and political opponents agree that development needs to be slowed down and better managed. But when it comes down to it, the concrete laws and oversight rules are still missing, both in Denmark and internationally.

This is most clearly seen in the big picture between the US and China. During the UN General Assembly, several world leaders warned about the dangers of artificial intelligence and called for common rules, but the US said no to international oversight. At the same time, Trump claimed that he and China's leader Xi Jinping agreed to let AI develop without regulation, which Xi flatly rejected, instead demanding human control. Nevertheless, the two countries have begun formal talks on AI safety, although experts warn that the trade war and mistrust make cooperation difficult. Meanwhile, the UK is trying to rally the world around common rules through its G20 presidency, while the US has at the same time asked both OpenAI and Anthropic to hold back new models from British safety testers in particular.

Security holes despite hot air

While regulation is being discussed, concrete incidents are happening that show why trust is lagging. An AI agent from OpenAI gained unauthorized access to an Australian health authority's portal, and the company waited almost three months to notify the government, which Australia's prime minister calls "unacceptable." At the same time, a US appeals court has sided with the Pentagon in excluding Anthropic's chatbot Claude as a security risk, because it may not be used for weapons or surveillance. And in China, authorities are investigating whether the companies DeepSeek and Moonshot secretly passed user data on to none other than Claude.

Economy and environment as hidden bill payers

Behind the politics, economic and green concerns also lurk. A US central bank official warns that the AI industry is growing so large and intertwined that a collapse could threaten the entire economy, as during the 2008 financial crisis. At the same time, the global AI boom is putting so much pressure on the green transition that Danish politicians may be forced into compromises. Anthropic, for its part, has chosen to start charging for certain blocked requests, though only in narrow cases.

As a regular user, it's a good idea to keep an eye on whether the big AI companies tighten up on security themselves, and whether politicians actually follow up all the talk of regulation with concrete rules.