Last week, the AI Act came a step closer. This European law aims to protect citizens from the dangers of artificial intelligence. But who is this legislation actually intended for? When will the act come into effect? What will you notice in your work? And the most important question: are you safe with AI?
Does the AI Act affect you?
Fellow author Tieme Worldman wrote about the AI Act a few days ago . I therefore no longer need to explain what this legislation entails. It is clear that the law is primarily intended for developers of AI applications. If you have tools built or set up, the act also applies to you.
The scope of the law also depends on what you do with AI, even if you use standard tools. The precise rules and standards still have to be worked out, but logically, applications in critical and high-risk sectors, such as medical care and infrastructure, fall more into the high-risk category.
Also logical: the law specifically mentions the use of AI in recruitment and personnel management as high risk. Bias can easily creep in there. That bias is a known problem with all artificial intelligence based on large language models , such as ChatGPT. That also immediately indicates the scale of the problem: the built-in prejudices are difficult to determine and filter out. It is not yet clear whether AI tools for customer experience management , for example , will also be classified as high risk because of this.
An urgent problem?
If AI tools or applications fall into the high-risk category, then all kinds of rules apply under the AI Act. Or rather: then all kinds of rules will apply. Although we are ahead in the EU, the development of legislation on AI is unfortunately much slower than the development of AI itself.
The legislative process has been ongoing since April 2021, when no one but a bunch of geeks had heard of Generative AI. Via the European Commission and the European Council, we have malta email list 149912 contact leads now arrived at the European Parliament, which will hopefully adopt the final version of the AI Act by the end of this year. Then the standards will have to be worked out and adopted, after which the law can hopefully enter into force in 2026 .
What should you do with it now?
OK, so you still have at least two years to get your act together. But it’s even better to do it now. First, the legislation is seen as sensible and reasonable – even, after some resistance, by how interactive content helps you achieve your marketing goals the AI developers themselves. Second, the use of AI is bound to take off in the near future. And then you might as well get it right the first time.
For you as communication specialists, this will probably also involve generative AI tools. The law specifically addresses this. If you use generative AI, then under the AI Act you must clearly state that the aleart news content is AI-generated, even if it concerns low-risk use.
I understand that when you talk about an autonomous helpdesk chatbot . For creative purposes, that of course raises questions: what if you have a draft generated by AI and adjust it yourself? What if you edit or combine pieces of original content with AI? These will have to be answered during the development of the act.
Also read: The AI Act is coming: here’s what you need to consider
What is useful is that generative AI apps will have to be open about copyrighted content that they use. That will also give you clarity as a business user.