EU's AI Act: Ethical and Transparent AI Regulations & Implications for the US

I received this email from Coursiv (an online AI educational platform)

(beginning email) Hello,

You know how we’ve been exploring AI tools and enjoying what they can do? There’s another side to that coin: making sure AI is used responsibly. I wanted to tell you about a major development that occurred in the European Union. In 2024, they passed a landmark law called the AI Act, which is basically the first big set of rules for AI anywhere in the world. Even if you’re not in Europe, this is likely to affect the AI products we all use.

Why a law for AI?
The EU’s law is all about safety, transparency, and trust. They want to encourage innovation in AI, but also protect people from potential harms like biased algorithms or creepy deepfakes.

Here are a few key things the AI Act does, in simple terms:

• • • AI has to introduce itself. If you’re chatting with a bot, the law says you have the right to know it’s a bot. No AI pretending to be human without telling you. So in the future, when we use a customer service chatbot or any AI agent, expect a little note saying “Hi, I’m an AI assistant”. This transparency is great - it builds trust because you know what you’re dealing with.
    • Labels on AI-generated content. Any AI-generated images, videos, or audio that could be mistaken for real should be clearly labeled as AI-made. So, if someone makes an AI video of a politician saying something they never said, legally it should have a watermark or warning. This helps combat misinformation.
    • Banning the scary stuff. The law bans certain uses of AI that are considered too dangerous. For example, AI systems that do “social scoring” - like rating your trustworthiness based on your behavior or personality - are not allowed. Also banned: AI that exploits people’s vulnerabilities (imagine a toy with AI urging a kid to do something harmful - not allowed).
    • Strict rules for high-risk AI. Some AI systems have big impacts on our lives - think about an AI that helps decide if you get a loan, or if your CV gets shortlisted, or an AI diagnosing your X-ray. The EU calls these “high-risk AI.” The law will require such systems to be thoroughly tested for accuracy, fairness, and bias, with human oversight in place. You should be informed and there should be a way to appeal or have a human review it.

I find it encouraging that governments are addressing the ethical and social impacts of AI. It means as we continue learning and using these tools, there’s an increasing safety net. It’s a big change, but a positive one to keep AI beneficial. (end email)

Lamps 3:16

The European Union's AI Act sets a precedent for responsible AI use, emphasizing safety, transparency, and trust through measures like mandatory AI disclosure and clear labeling of AI-generated content. These "sane and reasonable" regulations, particularly the classification of certain AI applications as "high-risk," could significantly influence the adoption of AI regulations in the United States.

If the US follows a similar trajectory, insurance companies, given their critical role in individuals' lives, are highly likely to be deemed "high-risk." This designation would subject them to intensive scrutiny, potentially leading to a more cautious and sparing use of AI to avoid regulatory and litigation challenges. Furthermore, the concept of "autonomous" insurers heavily reliant on AI could face immense pressure from regulatory bodies, potentially muting some of the operational advantages they might otherwise achieve.