ChatGPT is one of the most famous AI bots. However, new regulations are set to be introduced in the EU to regulate AIs. The founder and inventor explains that ChatGPT will be taken offline in Europe if these are too strict.
There have been discussions for some time about regulating AI systems. In Europe, there are already initial draft laws, but the head of Open AI, Sam Altman, is not particularly excited about it.
Altman is currently traveling around the world, speaking with large tech companies, businesses, and political decision-makers about AI models. He is specifically concerned about AI-friendly laws. Above all, the possible regulations from the EU do not make him happy, and there are various reasons for that.
ChatGPT will leave the EU if it cannot comply with the laws
Sam Altman, founder and inventor of OpenAI, spoke out against strong regulations on AI during a panel discussion in London. He would be disturbed if the EU defines AI systems like ChatGPT as “high-risk systems.”
Altman said that based on how the proposed law is currently formulated, both ChatGPT and the large language model GPT-4 could be classified as high-risk. Then the company would also have to implement new guidelines.
As the CEO of OpenAI explained (via gizmodo.com):
If we can meet the requirements, we will, and if not, we will shut down the operation [of ChatGPT]… We will try. But there are technical limits to what is possible.
The “problem” here is also the European Data Protection Board of the EU. They stated that they want to closely examine ChatGPT. The AI must comply with the EU’s data protection laws. In the USA, there were apparently no such major concerns; in the EU, they look a little closer.
EU AI Law was proposed back in 2021
The EU AI law is one of the laws that the EU Commission proposed in 2021. In this draft, AIs are classified into three risk categories (via europaparl.europa.eu).
The EU already has clear bans in mind. Systems are to be prohibited from the outset,
- that use subliminal or intentionally manipulative techniques.
- exploit the vulnerabilities of people.
- be used for social scoring (classifying people based on their social behavior, socioeconomic status, or personal characteristics).
In the lowest category are so-called “high-risk AI systems.” These are allowed but must adhere to certain standards for transparency and oversight.
And this point is a thorn in Altman’s side. He opposes the EU regulations requiring OpenAI to disclose the sources of training data.
The inventor of AI ChatGPT reveals which 34 professions an AI will never be able to replace
