Barbara Wimmer
Digital State Secretary Florian Tursky (ÖVP) demanded these days that artificial intelligence (AI) must be regulated quickly so that companies that work with AI and want to offer AI applications have legal certainty.
The EU works with the so-called “AI Act“ We have been working on such an AI regulation for 2 years. Unfortunately you had generative AI like ChatGPT, Midjourney & Co not on the radar, which is why it was now necessary to create separate passages in the AI Act for it. It took some time before agreement was reached here.
Tursky, the EU process is taking too long
Tursky takes all this too long. He has one open letter Written to the EU Commission in Brussels, asking for urgency. “A European AI regulation must come before the end of this year. Vestager is now putting pressure on himself and wants to come to an agreement in 2023, so my letter has had an effect,” said the Secretary of State for Digitization in response to futurezone’s inquiry. “I don’t understand why it’s taking so long. We have to react wisely now and regulate artificial intelligence sensibly, it would be wrong to build for the time being,” says Tursky.
The AI Act is currently on the table in the EU Parliament anyway. Im responsible for the AI Act EU parliamentary expulsion The draft regulation is to be voted on Thursday so that the trilogue negotiations between the member states, the Commission and Parliament can start in the autumn and the necessary EU-wide process can continue. This means that the AI Act is expected to come into force in 2025.
Not going it alone, but anticipating the AI Act
Meanwhile, Tursky would like to be in Austria ab 2024 introduce their own AI authority. But what is the purpose of going it alone? As came out in the futurezone discussion, there is no plan to go it alone, but only to take up the planned rules from the AI Act ahead of time. “As part of the implementation of the AI Act, national bodies for the Certification of high-risk applications have to be named. We already know that and a new authority must be created for this purpose,” says the Ministry of Finance, which houses the State Secretariat for Digitization.
According to the AI Act, different rules should apply to high-risk applications of AI than to AI that anyone can use. AI applications fall into four categories: minimal risk, low risk, high risk, and unacceptable risk. Subliminal manipulation by AI, the exploitation of vulnerable groups such as children, or the Real-time remote identification. This is about mass surveillance through face recognition meant, for which there should be exceptions.
office in the ministry
This requires an AI authority that will make such an assessment of which applications fall into which categories. “In Austria we are thinking of one here phased plan. In a first step, a Office in our resort set up, which takes care of the preparations and the national implementation,” says the ministry. “When the AI Act is then implemented, the office will become a separate authority,” it said.
Currently, many AI applications that are popular with users come from the USA. ChatGPT is about also in popular apps like Bing or Snapchat been integrated and thus access is very low-threshold. How are you going to get these corporations to play by the rules? “The regulation will apply to the entire EU market, which is why the operators of AI software must submit to this regulation if they want to participate in the European market,” said the press spokesman for the ministry.
