The code of practice sets out how to meet incoming EU rules on disclosing deepfakes
Nearly 190 companies signed up to a code of practice on AI transparency designed to help developers and others meet incoming EU rules for labelling highly realistic AI-generated content, the Commission revealed on Friday.
Large AI developers including Anthropic, OpenAI, Google, Meta and France’s Mistral are among the list of signatures published by the Commission.
But large Chinese AI developers are notably absent from the list – such as DeepSeek and Alibaba, which have significant usage in Europe. The only Chinese signatory is computer maker Lenovo.
On 2 August, rules in the EU’s AI Act targeting risks around AI deepfakes kick in. From Sunday, highly realistic AI-generated images, video or audio must be clearly labelled as synthetic to avoid falling foul of the law.
