The Election Commission of India (ECI) has issued a mandate requiring all political parties to label any AI-generated content they might be using during election campaigning. The notice comes days after the Aam Aadmi Party (AAP) faced a police report for allegedly posting AI-generated content of Prime Minister Modi and Home Minister Shah on its party’s X account.
The notice clearly states that parties are required to “Clearly label any image, video, audio or other materials generated or significantly altered by AI technologies with a notation such as “AI-Generated” / “Digitally Enhanced”/”Synthetic Content”.”
This isn’t the first time the ECI has had to step in to check AI-generated content generated by political parties, either. During the Indian General Elections in 2024, the ECI issued guidelines to political parties discouraging the use of AI-generated content and deepfakes in addition to any other tools that might be used on social media to spread misinformation. These guidelines put a blanket ban on any publishing or circulation of deepfake audio or video content from a party’s social media handles. If published, the content was to be taken down within three hours and a warning should be issued to the accountable person.
Other than AAP’s alleged use of AI-generated content, the BJP also pulled similar tactics. BJP’s national spokesperson Sudhanshu Trivedi revealed a set of audio recordings alleging opposition leaders, MPs, and police officers were collaborating to run illicit financial activities to fund the opposition’s election campaign on the eve of the Maharashtra state elections on November 19. The recordings were later found to be AI-generated by multiple news outlets and fact-checking organisations.
All major organisations in AI have started changing their models to include internal watermarks to tell AI-generated content apart. OpenAI and Meta, the two biggest companies in AI, have added signature watermarks to their respective image generators. However, while digital watermarking is a potential solution, it can be broken, provided an adversary has enough technical knowledge.
In the News: Global IoT botnet unleashes large-scale DDoS attacks
