Thursday, May 23rd 2024
European Union's AI Act Adopted, Full Implementation to Follow by 2026
The European Union has recently approved the AI Act, a new type of regulation to ensure the safe and responsible development of artificial intelligence (AI) within the EU. This legislation marks the first case in the global efforts to regulate AI, as it sets a new standard for transparency, accountability, and ethical considerations in AI development and deployment. The AI Act, also known as the KI-Gesetz, is designed to address concerns about the potential risks and negative impacts of AI on society. Key aspects of the regulation include the requirement for AI systems to be transparent and explainable, ensuring that they are not discriminatory and environmentally friendly. Furthermore, the AI Act emphasizes the importance of human oversight in AI development and deployment, ensuring that other technologies do not solely control AI systems.
The regulation also includes specific restrictions on certain AI applications, such as mass surveillance using biometric data or the evaluation of social behavior, similar to China's Social Scoring system. Additionally, emotion recognition, both in the workplace and educational settings, is prohibited under the new law. While the AI Act has been widely praised for its forward-thinking approach, there are also concerns about the potential limitations and challenges it may pose. The Digital Industry Association, Bitkom, has criticized the law for leaving certain questions unanswered and warning that the regulations may become outdated within two years. The AI Act is set to come into effect in the coming weeks, with full implementation expected by 2026. The regulation will apply to all entities developing, offering, or using AI systems within the EU, including both public and private parties. This comprehensive approach aims to ensure that AI is developed and used in a way that benefits society while minimizing potential risks.
Source:
HardwareLuxx.de
The regulation also includes specific restrictions on certain AI applications, such as mass surveillance using biometric data or the evaluation of social behavior, similar to China's Social Scoring system. Additionally, emotion recognition, both in the workplace and educational settings, is prohibited under the new law. While the AI Act has been widely praised for its forward-thinking approach, there are also concerns about the potential limitations and challenges it may pose. The Digital Industry Association, Bitkom, has criticized the law for leaving certain questions unanswered and warning that the regulations may become outdated within two years. The AI Act is set to come into effect in the coming weeks, with full implementation expected by 2026. The regulation will apply to all entities developing, offering, or using AI systems within the EU, including both public and private parties. This comprehensive approach aims to ensure that AI is developed and used in a way that benefits society while minimizing potential risks.
19 Comments on European Union's AI Act Adopted, Full Implementation to Follow by 2026
/s
I'm not sure if EU gets the ridiculousness of trying to regulate domain where we aren't leading by any measure. Create the prerequisites to be at the front, lead in some way and then you can think of regulating.
And I don't mean users shouldn't be protected. But the way EU does things, the net result is everything noteworthy happens elsewhere and we just consume (i.e. generate revenue for non-UE entities) that.
Seems like it's working quite nicely
No compliance? Good, GTFO!
And what's more, people are actually starting to see its value too.Not sure, I think the EU is way ahead of the inevitable reality check that will happen and suddenly it all makes sense to everyone. Its guidance by principle, which I personally think is the only way to write law and policy. You're writing it to protect those principles.
Something AI is going to fail spectacularly and make casualties, sooner or later. Or, its just not going to go places altogether and we don't really need the regulation.
Again, look at GDPR. The EU has no social media of its own, it hits all companies and we're not leading in tech, but companies must get in line or they lose access to a rough third of the world's economical power. They ain't leaving.
Isn't that a separate issue? We've been neglecting a lot of things in that space.
The reason isn't regulation. We ourselves outsourced things.
Anything noteworthy happens elsewhere... such as this?
www.asml.com/en
Or this?
www.nxp.com/
qt.eu/
Or this?
bellona.org/news/eu/2021-03-hydrogen-in-steel-production-what-is-happening-in-europe-part-one
Unless... of course, we consider 'noteworthy' the Big Tech clusterfuck of social media and information control/overload, over the above.
You know what, let those things happen outside the EU. We'll focus on stuff that matters ;)
To be fair, Europe is in the process of hitting a wall with considerable speed when it comes to social policies - in no small part thanks to intellectually weak people being in charge of immigration - but in the technology realm, "quality over quantity" seems to be the chosen trajectory. Doing better things in a better way might in the end win with churning chinesium rubbish and a race to the bottom with price and quality, which seems to be exactly what the USA is doing.
One of the biggest sources of revenue these days is services: you can build them once and sell them ad nauseam at very little cost per "SKU". No mainstream social media platform is based in the EU. AI? Anywhere but EU. Cloud providers? Same answer.
And if you want to talk physical goods, pretty much everyone banned that (because green and costs) and moved production to China or India.
You can rationalize that all you want, but at the end of the day, it means people willing to work on bleeding edge stuff just moves elsewhere and we hurt in the long run.
So it's just there for the sake of it, not that somebody would take it seriously.
Too much money going out of the EU for Windows licenses? Fine Microsoft and get some of that back. Too much money headed towards Intel? Same thing. I expect this will be no different.
And don't get me wrong, I know full well AI needs to be at least properly supervised. I just don't see the EU making a good job of it.