Election Commission proposes AI rules in code of conduct for elections

The Election Commission of Pakistan has decided to formally use artificial intelligence (AI) in the electoral process. According to Dunya News, as cited by Daily Pakistan, it has proposed AI-related amendments to the codes of conduct for political parties, media and election observers.

The commission has sought suggestions from the public and political parties on preventing the misuse of AI in elections. The proposed AI code has been issued to political parties. The amendments revise the code of conduct the commission prepared in 2023, and suggestions are being accepted until 20 November.

For political parties, candidates and supporters, the proposal says election material created or altered with AI cannot be published without a clear label. Deepfakes, synthetic voices and videos must be clearly marked as AI-generated. Creating or publishing AI content that deceives voters would be prohibited.

The draft also says that falsely creating a person's face, voice or statement would be strictly forbidden. Fake deepfake content targeting judges, election officials and the armed forces would be banned. Using AI bots and automated accounts to create fake trends and likes would also be barred.

Robocalls and voice-clone messages would have to state at the start that they are AI-generated. Parties would have to keep a record of all AI content and provide it to the commission.

For the media, the proposal requires print, electronic and digital outlets and social media influencers to label AI content clearly. Anyone airing AI-made images, video or audio about a candidate or election matter would have to say so. Fake, insulting and defamatory deepfakes would be strictly prohibited.

Media houses could use AI only for content generation, translation and analysis. PEMRA, the SMPRA and the NCCIA would monitor AI content, and action and reports on unlabelled content would be sent to the commission. On the commission's instruction, media would have to remove misleading AI content immediately or add a correct label.

A separate proposed code covers election observers. They would watch for deepfakes and synthetic content, report suspicious AI material to the commission at once without sharing it, and not alter their own observations, photos or videos with AI tools.