The European Union (EU) has announced that its new rules on the transparency of AI systems have now taken effect and are enforceable across the union as of August 2nd.
Under the new rules, chatbots and other interactive AI systems will have to notify users that they are dealing with AI, and not a human. Alongside this, images, videos, or audio that have been edited or created using AI will have to be labelled. This AI-generated or even AI-altered content will also have to carry machine-readable marks so it can be more easily detected.
The EU notes that the measures are “intended to reduce deception and manipulation and help people make informed choices,” according to the press release. “They also give businesses clearer obligations and a practical way to show compliance.”
Also included in the new rules are general-purpose AI models. Those organizations will now be required to document information, publish training data summaries, and implement copyright policies, with non-compliance of these rules subject to enforcement by national authorities.
As Engadget pointed out, the new laws carry some pretty major penalties, with fines for companies ranging up to €15 million or three per cent of global annual revenue, and fines for institutions, bodies, and agencies of €750k.
Interestingly, the rules don’t include personal content like group chats, and “evidently artistic” satirical and fictional works.
These new rules come just as Statistics Canada released a survey noting that one in three Canadian workers used generative AI, and as Mississauga approved a one-year pause on AI data centres.
Source: The European Commission via Engadget
Image Credit: Shutterstock
