EU: how effective will the world's first AI law be?
Negotiators from the EU Parliament and member states have agreed on the key points of the AI Act which has been in preparation since 2021. The law aims to guarantee transparency on the use of artificial intelligence, ensure high quality of the data used in development and protect copyrights. In the areas of data protection and security-relevant applications, human-controlled risk management will become mandatory.
An important and reliable framework
Finally Brussels has passed legislation that won't kill innovation, Handelsblatt writes in delight:
“Of course the regulation imposes requirements on all companies wanting to develop basic AI models. Anyone who uses vast amounts of data to train models that can process and generate texts, images, videos and programming code must be transparent about how they go about it. But that is a demand that at least corporate customers should be making anyway. .... The EU has created a reliable framework for the use of AI - now it needs to invest in development and infrastructure. This is crucial on the path to technological sovereignty.”
Right direction, key exceptions
The Irish Examiner welcomes the act:
“In this weekend's agreement, the European Parliament secured a ban on use of real-time surveillance and biometric technologies, including emotional recognition, but with three exceptions: Police would be able to use the intrusive systems only in the event of an unexpected threat of a terrorist attack; the need to search for victims; and in the prosecution of serious crime. ... The digital road ahead will be long and complicated, but Europeans have now taken the first steps along it.”
Liability issues remain unresolved
The AI Act leaves some crucial questions unanswered, comments Kapital:
“The question of who is responsible for any damage caused by AI, for example. If an autonomous car has an accident, who pays for the damage? The owner, the manufacturer or the software developer? If the manufacturer is completely exempt from liability they have no incentive to offer a good product or service, which could weaken confidence in the technology.”
Champions at legal protection, weak on the market
Unfortunately the EU is not as strong on other fronts as it is in regulation, La Stampa chides:
“So far, Europe has always won the race for rules to protect citizens, but has always lost on the market. The giants of the internet are almost all American. ... When it comes to artificial intelligence, the script hasn't changed. The US and China are leading the race, Europe is lagging behind. The gap on the microchip front, the processors required for artificial intelligence, is glaring. The European Chips Act, passed a few months ago aims to close this gap by doubling our production by 2030. But there's a long way to go before that happens.”
More education, fewer restrictions
For El País the regulation is good, but not good enough:
“The alarm artificial intelligence has caused within society is almost entirely based on its aberrant uses, such as the generation of false reports or fake nude pictures. ... Education would do more to resolve this problem than a law that restricts technical advances. Blaming artificial intelligence for everything is an evasive stance. And the plague of disinformation will not disappear no matter how many cyber regulations Europe passes. ... When it comes to science and technology, legislators are condemned to always lag behind while trying to keep up.”