EU Considers Stricter Rules for Big AI Models: Report

Article source

In the European Union, negotiators are currently exploring the possibility of imposing additional restrictions on sizable AI models, including OpenAI’s ChatGPT-4, as part of the forthcoming AI Act.

According to a report from Bloomberg, representatives within the European Union are engaged in discussions concerning the potential impact of large language models (LLMs), which encompass Meta’s Llama 2 and OpenAI’s ChatGPT-4.

They are deliberating the prospect of introducing further regulatory measures as an integral facet of the impending AI Act.

Sources familiar with the matter, as reported by Bloomberg, convey that the overarching objective is not to overwhelm nascent startups with an excess of regulations while maintaining oversight over larger models.

It is worth noting that the consensus reached by negotiators on this issue is still in its nascent stages.

The AI Act, coupled with the proposed regulations for LLMs, represents a similar approach to the European Union’s Digital Services Act (DSA).

Recently, EU lawmakers enacted the DSA, which entails establishing standards to safeguard user data and detect illicit activities on platforms and websites.

Nevertheless, it’s important to underscore that the most prominent online platforms are subject to more stringent controls.

Companies in this category, such as Alphabet and Meta, were required to align their service practices with the new EU standards by August 28th.

The EU’s AI Act is poised to be one of the initial sets of compulsory rules for AI instated by a Western government.

China, on the other hand, has already implemented its own set of AI regulations, which took effect in August 2023.

Under the EU’s AI regulations, companies involved in the development and deployment of AI systems would be obliged to conduct risk assessments, label AI-generated content, and would be entirely prohibited from utilising biometric surveillance, among other measures.

It is essential to bear in mind that the legislation has not yet been formally enacted, and member states retain the prerogative to voice dissent on any of the propositions put forth by the parliament.

Notably, since the implementation of its AI laws, China has reported the release of over 70 new AI models into the market.