In a move aimed at enhancing transparency, the European Commission is urging major online platforms to implement clear labeling for content, including text, images, and audio, that is generated by artificial intelligence (AI). Vera Jourova, the EU commissioner for transparency, called upon the 44 companies and organizations that have signed the bloc's voluntary code of practice to combat disinformation to incorporate this labeling requirement into the agreement.
Jourova emphasized the need for signatories, such as Microsoft's Bing Chat and Google's Bard, to integrate adequate safeguards into their AI services to prevent malicious actors from leveraging them for disseminating disinformation. Additionally, companies that offer services capable of propagating AI-generated disinformation should develop technology to detect and prominently label such content for users.
Microsoft and Google, along with platforms like TikTok, YouTube, and Meta (formerly Facebook), comply with the EU's code of practice. However, Twitter recently announced its withdrawal from the agreement, despite certain aspects of the code on disinformation being slated for inclusion in the forthcoming Digital Services Act, which will be enforceable by law starting August 25.
Since the emergence of ChatGPT, a popular chatbot, generative AI content has seen a significant surge. Notable examples include realistic yet fake images, such as the pope wearing a high-fashion puffer jacket and fabricated scenarios like the arrest of former US President Donald Trump. Deepfake videos featuring celebrities like Tom Cruise and music tracks that closely mimic popular artists such as Drake and The Beatles have also proliferated.
Industry experts anticipate AI's increasing role in interactive customer interactions via phone and computer, its simulation of famous actors in movies and commercials, language instruction, and more.
Jourova stressed the need for code-compliant companies to swiftly implement AI labeling so that "ordinary users can easily discern between content generated by humans and that produced by AI systems." The European Union is currently developing draft legislation to regulate AI, with specific provisions focusing on generative AI. However, negotiations between EU lawmakers and member states have caused delays, and even if the law is adopted by the end of this year, it would not come into effect until at least late 2025, according to EU internal market commissioner Thierry Breton.
