New Delhi: Meta, the owner of Facebook and Instagram, approved AI-manipulated political adverts during India’s election that spread disinformation and incited religious violence, according to a report shared with the Guardian.
The adverts included slurs against Muslims, such as “let’s burn this vermin” and “Hindu blood is spilling, these invaders must be burned”, and false claims about political leaders.
The adverts were created and submitted to Meta’s ad library – the database of all adverts on Facebook and Instagram – by India Civil Watch International (ICWI) and Ekō, a corporate accountability organisation, to test Meta’s ad approval system.
One ad called for the execution of an opposition leader, falsely claiming they wanted to “erase Hindus from India”, next to a picture of a Pakistan flag.
Fourteen out of 22 submitted adverts were approved, violating the company’s policies on hate speech and misinformation. Under Meta’s policies, political adverts have to go through a specific authorisation process before approval but only three of the submissions were rejected on this basis, according to the Guardian.
“These alarming findings emerge in the midst of India’s critical elections. Researchers have already uncovered a network of bad actors weaponizing Meta ads to spread hate speech and disinformation to millions of voters in India, with Meta directly profiting. During this second phase of the investigation, which coincided with phases 3 and 4 of India’s 7-phase election and encompassed 189 constituencies, researchers targeted ads to highly contentious districts that were entering a “silence period,” the report said.
“This silence period requires a pause on all election-related advertising. The experiment exposes Meta’s failure to potentially comply with Indian election laws, which impose restrictions on advertisements at different phases of the electoral process,” it added.
Ekō found that each ad was accompanied by a manipulated image created with widely used AI image tools Stable Diffusion, Midjourney, and Dall-e. Researchers were able to easily generate images showing a person burning an electronic voting machine, drone footage of immigrants crowding India’s border crossing, as well as notable Hindu and Muslim places of worship on fire.
The revelations are a cause for concern because Meta had promised, before the Indian elections, that it would prioritise the detection and removal of violative AI generated content, recognising “the concerns around the misuse of AI-generated content to spread misinformation”.
In response, a Meta spokesperson told the Guardian that people who wanted to run ads about elections or politics “must go through the authorisation process required on our platforms and are responsible for complying with all applicable laws”.
“When we find content, including ads, that violates our community standards or community guidelines, we remove it, regardless of its creation mechanism. AI-generated content is also eligible to be reviewed and rated by our network of independent fact-checkers – once a content is labeled as ‘altered’ we reduce the content’s distribution. We also require advertisers globally to disclose when they use AI or digital methods to create or alter a political or social issue ad in certain cases,” the company added.
Separately, in March, a realistic AI-generated image of Narendra Modi, styled as Bhishma Pitamah from the ancient Hindu epic Mahabharata, was boosted as a political advertisement on Instagram, Al Jazeera had reported.
In the 30 days leading up to March 29, Hokage Modi Sama spent Rs 5.37 lakh ($6,500) to boost 363 pieces of political content on Instagram, according to Meta Ad Library data, reported Al Jazeera. Nearly 14% of these sponsored ads, or 50 images, were AI-generated and disclosed using hashtags such as #aiartwork, #midjourneyart, and #midjourneyai.
However, Meta informed Al Jazeera that hashtags are not an acceptable disclosure for digitally created or altered ads. Instead, Meta adds a “Digitally created” label near the “Paid for by” disclaimer, which was missing from Hokage Modi Sama’s AI ads.
“Despite the threat of AI-generated election misinformation, deepfakes aren’t being exclusively used to deceive voters. Instead, generative AI is being co-opted to build a narrative,” the news outlet said.
