In an effort to mitigate transparency concerns surrounding chatbots and AI deepfakes, the European Union has introduced new guidelines requiring businesses to disclose their usage of AI models. These requirements, which are part of the EU’s transformative AI Act, came into effect on August 2, necessitating companies to clarify whether humans are interacting with AI models or whether the content they are consuming has been generated or modified by these AI models.
The directives vary between “providers” – companies that develop and sell AI systems – and “deployers” – platforms and services that use the AI systems. Interestingly, some firms, including the likes of Meta and SpaceXAI, find themselves classified as both. This division recognizes the diverse roles companies can play in the AI landscape and establishes regulations appropriately.
In addition to the requirements set for businesses, the European Union has also provided AI labels for companies to use. These labels come as a standardized solution, intended to be used across all platforms for the common goal of transparency and can be adopted instead of creating unique, company-specific ones. It is hoped that if enough companies adopt these labels, they will become a consistent symbol across online platforms, further fostering a uniform understanding for users in their interactions with AI systems.
One stipulation for providers under the new rules specifies that AI systems must be designed in a way that makes it explicitly clear when a user is interacting with such a system. This, once again, falls into the wider objective of promoting transparency and trust between users and AI technologies.
While the true effects of these additional transparency obligations will only become evident with time, they represent an important step forward in the regulation and control of AI, addressing growing concerns about transparency in the industry.
Credit: The Verge.
