Online campaigning: Urgent vigilance needed over AI-generated videos

As the 13th national parliamentary election draws closer, confusion and anxiety surrounding campaigning on social media are intensifying. The use of artificial intelligence to create fake videos or deepfakes, as well as the cloning of newspaper photo cards using inexpensive software to circulate false statements or chipfakes, has increased alarmingly.

We are witnessing deepfakes and chipfakes being weaponised as tactics to spread falsehoods and disinformation against rival parties and candidates. At the same time, AI-generated, unlabelled videos are also being used for campaigning on behalf of various parties and candidates.

In the context of Bangladesh, where digital literacy remains extremely low, heightened caution and responsible conduct by political parties in online campaigning are more essential than ever. How far this responsibility is actually being observed remains an open question.

An analysis by the fact-checking organisation Dismislab reveals that, ahead of the election, political campaigning using AI-generated videos is taking place across social media platforms such as Facebook, YouTube and TikTok. In these videos, no disclosure is made as to whether the content is real or created using artificial intelligence.

Between 1 and 15 January, the organisation identified 800 such videos on Facebook, 60 per cent of which carried no labels. Of the 181 videos found on YouTube, 94 per cent did not mention that they were AI-generated. Not a single one of the 50 videos identified on TikTok carried an AI label.

Recent elections in countries including the United States, India, Argentina and Turkey demonstrate how highly realistic AI-generated videos can shake voters’ trust and influence public opinion. Beyond physical violence on the ground, this form of digital violence has emerged as one of the greatest threats to modern democracy. In Bangladesh’s deeply polarised political landscape, this threat is even more acute.

The Representation of the People Order (RPO) clearly states that the use of artificial intelligence in elections to create, publish, disseminate or share false, misleading, biased, hateful, obscene, indecent or defamatory content is prohibited. The election code of conduct also explicitly bans the use of AI to spread misleading and false information.

We believe that the responsibility of the election commission does not end with merely amending laws. Despite clear violations of the election code of conduct on social media, no scientifically grounded or effective measures have so far been observed in the fight against disinformation.

Moreover, on Saturday, a serious flaw in an online system launched by the election commission (EC) led to the exposure of the personal data of at least 14,000 journalists. This has raised further questions about how vigilant and responsible the EC is regarding citizens’ data protection, digital security and personal privacy.

Information technology experts believe that deepfakes are not the only threat in elections; the risk of cyberattacks also demands the highest level of caution. On social media, many influencers are spreading false, misleading and hateful statements. What is particularly concerning is that, in many cases, such statements are fuelling social division.

We believe that the election commission must remain maximally alert and vigilant to ensure that AI-generated videos and misleading statements on social media do not, in any way, negatively affect the electoral environment. Political parties and candidates must also act responsibly in online campaigning and strictly adhere to the election code of conduct.