Key Takeaways:
– The European Union’s AI Act went into effect on August 1 and governs the use of AI applications across various fronts.
– Companies must comply with different requirements based on the risk level of the AI applications they’re implementing.
– The AI Act will significantly affect firms outside of Europe, particularly American AI leaders like Google, Meta and others.
– The legislation does not create specific regulations for generative AI, leaving an area for businesses to exploit.
– There are concerns about the Act’s potential impacts on innovation and its imposed compliance risks.
The Newly Enacted European Union AI Act: Questions Galore
On August 1, the European Union implemented the Artificial Intelligence Act — one of the world’s first laws designed to broadly regulate AI applications. The law, initially proposed to the European Commission in 2021 and fully approved by the European Parliament in March this year, aims to harmonize existing AI laws among EU members while presenting new legal frameworks for AI use.
Understanding the Framework of the AI Act
By seeking a unified regulatory framework, the AI Act is set to influence AI application development, companies’ use of AI, and the penalties resulting from law-breaking activities. The focus, here, is not on regulating specific technology but rather on the applications of AI, judging them primarily on the risks they pose.
AI applications fall under different risk categories. The lowest risk AI uses, such as search engines, face minimal regulation, while systems posing limited risks, like chatbots, are also legal but must comply with certain transparency requirements. For higher-risk applications, such as credit scoring or school admissions, companies need to secure approval. The law outright bans public facial recognition systems as they pose an unacceptable risk. Firms failing to adhere to these requirements face fines of up to €35 million or 7% of their worldwide revenue.
Addressing the Developed Inquiries about the AI Act
Being effective for less than a week, the EU AI Act raised many questions. Here are five critical ones addressed:
1. Who Will the AI Act Impact?
Like its precursor, the GDPR, the AI Act will have a substantial impact outside Europe. Predominantly, American firms like Google, Meta, OpenAI, Amazon, Apple, Anthropic, and others that lead the AI field are expected to experience this impact. These firms will have to be particularly careful about their data collection routines to prevent accidental capturing of European data.
2. What’s the Act’s Effect on Generative AI?
The AI Act doesn’t set out specific regulations for generative AI but bases its regulations on the use of AI. This is a positive note for companies aiming to utilize large language models and other foundational models. But companies are still required to meet some obligations when deploying GenAI applications, including disclosing how these models are trained and ensuring private data isn’t shared.
3. Will the Act Improve Life Quality?
While the EU lawmakers aim to prevent AI from causing harm through the new law, AI can also have beneficial impacts on life. Both aspects need balance, which brings up the question: have the Europeans found it? AI compliance firm CEO Guru Sethupathy used the evolution of motor vehicle regulation as an analog, emphasizing that smart regulations combined with technology will be crucial for building trust in AI systems.
4. Will the Act Stifle Innovation?
Though regulation risks choking innovation, it can spur adoption, thereby driving innovation as evidenced in the automotive industry’s history. The substantial question is the direction the AI Act will take, to which law expert Denas Grybauskas replies, the answer isn’t clear yet.
5. How Will the Act Influence American Regulations?
If GDPR sets any precedent, the AI Act will generate a lot of discussion, leading to laws at the state level, like the CCPA in California. Eric Loeb, executive vice president at Salesforce, suggests that the AI Act’s rules should be considered by other governments when crafting their policy frameworks.
