Since its release two months ago, ChatGPT has captured the imaginations of millions across the globe. The continued development and use of the tool, as well as those of other generative AI tools, might be banned under the EU’s AI Act.
With its AI Act, the European Commission is proposing the first-ever legal framework on AI and hopes to address its risks and position itself as a global leader on the topic.
While this framework will help regulate AI tools, it is also expected to have repercussions for generative AI tools such as ChatGPT as well as widely used tools such as Grammarly or GitHub Copilot, says think tank Center for Data Innovation.
According to EU industry chief Thierry Breton, these new proposed AI rules will curtail the risks – such as plagiarism, fraud and misinformation – posed by tools such as ChatGPT. While it’s easy to paint such rules as too radical, Breton is quick to point out the benefits of AI solutions.
“As showcased by ChatGPT, AI solutions can offer great opportunities for businesses and citizens, but can also pose risks. This is why we need a solid regulatory framework to ensure trustworthy AI based on high-quality data,” Breton told Reuters.
Under the current AI Act draft, ChatGPT is considered a general-purpose AI system which can be included for different purposes including those it deems “high risk”: such as the selection of candidates for jobs and credit scoring. Among others, the Act considers AI systems as “high risk” if they could put the life and health of citizens at risk; determine access to education and courses; are used in law enforcement or in the administration of justice and democratic processes. The other categories are “unacceptable risk”, “limited risk” and “minimal or no risk”.
AI companies are worried that their solutions may be placed in the “high risk” category as this would subject them to strict obligations before they can be put on the market. These include an adequate risk assessment, logging of activity to ensure traceability of results and a high level of robustness, security and accuracy.
The European Commission is working closely with the EU Council and the Parliament to further clarify the rules in the AI Act. Other aspects that still need to be discussed with lawmakers will include copyright violations, transparency and the risk of bias and false information.
