EU: ChatGPT spurs debate about AI regulation

Garante, the Italian data protection authority, apparently jumped the gun at the end of March when it imposed a temporary ban on ChatGPT, a chatbot that uses artificial intelligence (AI) to generate texts that seem as if they were created by humans, and computer games. The watchdog was less concerned by the use of AI — the simulation of human intelligence by computer systems — than by breaches of data protection legislation.

Garante then told the Microsoft Corp-backed company behind ChatGPT, OpenAI, that it would have to be more transparent with its users about how their data were processed. It also said that the US company had to obtain permission from users if their data were to be used to further develop the software — that is, to help it learn — and that access to minors had to be filtered. In a press release, the Italian authority said that the ban would be lifted if OpenAI met these conditions by April 30.

An OpenAI spokesperson told the Reuters news agency that it was "happy" that Garante was "reconsidering" the original ban and that it looked forward "to working with them to make ChatGPT available to our customers in Italy again soon."

EU-wide regulation of AI

Spain and France have also raised similar concerns about ChatGPT. For the moment, there is no EU-wide regulation of the use of AI in products such as self-driving cars, medical technology, or surveillance systems. The European Parliament is still debating legislation proposed by the European Commission two years ago. When it is approved, the EU member states themselves will have to agree, and so it will probably be early 2025 before it comes into force.

However, German MEP Axel Voss, one of the main drafters of the EU's Artificial Intelligence Act, pointed out that AI was not so advanced two years ago and was likely to develop further over the next two years, "so fast" that much of it would no longer be appropriate when the law actually took effect.

It is not clear whether ChatGPT or a similar product would even be covered by the EU regulation, which defines levels of risk in AI that run from "unacceptable" to "minimal or no risk." As the legislation stands, only programs assigned scores of "high risk" or "limited risk" will be subject to special rules regarding the documentation of algorithms, transparency and the disclosure of data use. Applications that document and evaluate people's social behavior to predict certain actions will be banned, as will social scoring by governments and certain facial recognition technologies.

Legislators are still discussing to what extent AI should be allowed to record or simulate emotions, as well as how to assign categories of risk.

Voss said that "for competitive reasons and because we are already behind, we actually need more optimism to deal with AI more intensively. But what is happening in the European Parliament is that most people are being guided by fear and concerns and trying to rule out everything." He added that the EU members' data protection commissioners wanted AI to be monitored by an independent body and that it would make sense to amend the existing data protection legislation.

Striking a balance between consumer protection and economy

The European Commission and Parliament are trying to strike a balance between consumer protection, regulation and the free development of the economy and research. After all, as the EU Commissioner for the Internal Market Thierry Breton has pointed out, AI offers "immense potential" in a digital society and economy. Two years ago, when the bloc's AI legislation was presented, he said that the EU did not want to drive the developers of AI away but promote them and persuade them to settle in Europe. He added that the EU should not be dependent on foreign providers and that the data for AI should be stored and processed in the EU.

Mark Brakel from the US-based nonprofit Future of Life Institute told DW that companies also had to be held accountable by regulators. He said that it did not suffice to apply risk levels to AI applications. He suggested that developers themselves should have to monitor the risks of each individual application and that measures should be taken to ensure that "companies are mandated to do this risk management and publish" the results. He added that sometimes companies could not predict today what their AI products might be able to do tomorrow and were sometimes surprised by the results.

"If we are too complicated here, then companies will go elsewhere and develop their algorithms and systems there," warned MEP Voss. "Then they will come back and use us only as a consumer country, so to speak."

What is striking about ChatGPT, which is causing a stir in Europe, is that it was developed in the US for global use. OpenAI could soon face stiff competition from other US companies such as Google and Elon Musk's Twitter. Chinese tech giants are also in the race, with Baidu already having created a chatbot called Ernie.

So far there do not seem to be any European chatbots on the horizon.

This article was translated from German.