𝗘𝗨'𝘀 𝗔𝗜 𝗔𝗰𝘁 𝗔𝗽𝗽𝗿𝗼𝘃𝗮𝗹 as world's first comprehensive AI law

March 13th, 2024 EU AI Act is Approved. 🌍 Let's think about this.

Hey Everyone,

I guess it’s worth mentioning that The European Union’s parliament on Wednesday, March 13th, 2024, approved the world’s first major set of regulatory ground rules to govern the mediatized artificial intelligence at the forefront of tech investment.

These new rules will take effect in stages between late 2023 and mid-2026 across the 27 EU member states.

European Union leaders have taken the lead over much of the rest of the world on setting AI policy, forging an agreement to regulate the groundbreaking tech after more than 37 hours of negotiations last week.

Clearly a lot will happen between now and 2026. However, after three years of work, 523 European Union lawmakers voted for the proposed agreement and just 49 voted against it (with 49 abstaining). Formal approval is now set to follow in March. Legislation will then begin early next year and apply from 2026.

However whether Europe will even be relevant in the age of Generative AI is another question altogether. Outside of Paris France it doesn’t seem like they will be able to keep up with either the U.S. or China.

On Wednesday, Parliament approved the Artificial Intelligence Act that ensures safety and compliance with fundamental rights, while boosting innovation.

It’s not clear if this will boost innovation. Born in 2021, the EU AI Act divides the technology into categories of risk, ranging from “unacceptable” — which would see the technology banned — to high, medium and low hazard.

What will AI even be by 2026?

The 2026 'rulebook' will bring plenty of potential overhead and compliance burdens like bans / fines. For certain fields the costs of compliance could be significant.
It’s still not clear how this deals with fundamental X-risks posed by AGI like systems. The most concerning are the Act's provisions around extremely powerful AI that could present systemic risks.

The regulation is expected to enter into force at the end of the legislature in May, after passing final checks and receiving endorsement from the European Council. It’s not clear how A.I. will evolve in the two years before this actually takes effect.

“Europe is NOW a global standard-setter in AI,” Thierry Breton, the European commissioner for internal market, wrote on X. The AI Act works by classifying products according to risk and adjusting scrutiny accordingly. While OpenAI’s Sora debates if their text-to-video should include nudes, there’s clearly a bifurcation around AI regulation between the U.S. and Europe.

The EU AI Act will try to ban specific applications and some real-time AI like remote biometric identification, trying to predict whether someone will commit a crime (that reminds me of minority report), classifying people for social-scoring purposes, inferring people's emotions in workplace/educational institutions and more.

Europeans seem to think that this is a big deal, I am not so sure. The law will allow for facial and biometric scanning, though there would be limits and exceptions. Powerful American and Chinese companies will continue to lobby to find their own ways around some of the rules.

An Internet of Synthetic Nudes - Sora'