Munich AI accord

Big Tech commits itself to countering election misuse

On Friday, the world’s 20 largest tech companies and social-media platforms signed an accord, “The Tech Accord to Combat Deceptive Use of AI in 2024 Elections”, at the Munich Security Conference. This is a commitment to prevent deceptive artificial-intelligence (AI) content from interfering in elections. It is timely — over 40 nations, including India, the US, and the UK, are slated to vote in new governments in 2024, affecting the fate of over four billion people. Harmful AI-generated content meant to deceive voters could cast a malignant shadow over democratic processes. The signatories will collaborate to detect and address online distribution of fake AI content, drive educational campaigns, and provide transparency about AI usage in generating political content. The accord also includes a broad set of principles, such as the importance of tracking the origin of deceptive political content and the need to raise public awareness about it. Using modern tools, it is easy to generate authentic-seeming audio visual (AV) content featuring a politician, or indeed anybody.

The use of AI as such may not be harmful to democratic processes. Campaigns legitimately using AI should label and flag it, and it should not be deceptively deployed. Imran Khan’s political party, the Pakistan Tehreek-e-Insaf (PTI), used AI to generate speeches penned by him during the recent Pakistan elections and this allowed him to campaign from behind bars. However, AI was also used in the recent New Hampshire (US) presidential primaries to generate fake robocalls in Joe Biden’s voice to dissuade voters from turning up. The content the accord addresses could be AI-generated audio, video, or images that fake or alter the appearance, voice, or actions of political candidates and election officials, or provide false information about voting processes. This could be done to discredit a politician, or distort a political message, or confuse voters and dissuade them from exercising their franchise.

The accord aims to develop methods to identify deceptive content, label it, and prevent dissemination on their platforms. Meta, X, Google, LinkedIn, IBM, Adobe, OpenAI, Amazon, TikTok, and Microsoft are signatories. A successful collaborative effort by them would restrict the spread of such content on large social media platforms or search engines. Moreover, many key players developing AI tools, such as OpenAI, are also signatories. They could try to monitor use of AI tools, or deny access to these for generating political content.

But such tools are also evolving. It will not be an easy task to prevent AI usage or detect deceptive content, and check it from going viral. It is unclear how this is to be accomplished. It will take a big research and development effort by the industry to detect and label AI usage in political AV content, and to discover the originators of such stuff. Moderating it would also be a huge task. Lawmakers would have to cooperate, and some may be reluctant to do so if the content concerned aids their cause. Moreover, it needs to be pointed out that a “bad actor” using AI to make political content need not necessarily be hired by a political party. It could be an overseas agency pushing the agenda of another nation. Or, it may be somebody trying to create engagement, using political content as “clickbait”. It is important that this campaign is not merely symbolic due to the high stakes. If the signatories don’t get it right, democratic processes could indeed be subverted with terrible global outcomes.