See question(s) : E-005183/2021
EN
E-005183/2021
Answer given by Mr Breton on behalf of the European Commission
(17.1.2022)
The recent Commission proposal for a Regulation laying down harmonised rules for Artificial Intelligence (AI)1 defines ‘deepfakes’ as generated or manipulated image, audio or video content produced by an AI system that appreciably resembles existing persons, objects, places or other entities or events and would falsely appear to a person to be authentic or truthful (‘deep fake’). Under the proposal, deepfakes will need to be labelled unless their use is authorised by law for law enforcement purposes or for the exercise of right to freedom of expression and of the arts and sciences.
In addition, the EU also tackles certain aspects related to deepfakes through the Code of Practice on Disinformation, which was created in 2018 by major online platforms and stakeholders from the advertising industry. The Commission issued Guidance in May 2021 on how to strengthen the Code2, which foresees that signatories should commit to take action against novel and emerging techniques of disinformation, including deepfakes, and should establish a permanent task force to ensure that actions against all forms of disinformation remain timely and relevant.
Finally, it is also worth noting that already today, deepfakes are not unregulated. Deepfakes may be considered unlawful under existing civil or criminal law, for instance because they involve impersonation, defamation, fraud, incitement of hatred or other crimes or unlawful processing of personal data for the generation/manipulation of the content.