Facebook Flooded with Fabricated AI Content: PIB Report Reveals Rising Disinformation Trend

Artificial Intelligence (AI)-generated content is increasingly dominating social media — and with it, a surge in disinformation. Video content, in particular, has emerged as the leading medium for spreading such misleading materials.

A recent analysis of 71 AI-generated photos and videos circulating on social media over the past three months found that 57 were video-based while 14 were image-based. According to findings by BanglaFact, the fact-checking and media research wing of the Press Institute of Bangladesh (PIB), Facebook topped the list as the primary platform for disseminating AI-based disinformation, accounting for 86% of Bangladesh-related misleading content.

Following Facebook, TikTok ranked second, where around 56% (40 items) of such AI-generated falsehoods were detected. The BanglaFact team also identified 32 disinformation cases on YouTube, 28 on Instagram, 12 on X (formerly Twitter), and 1 on Threads.

The report observed that the most viral content involved sensational and politically charged topics, with AI technology being weaponized for harassment and defamation. It also noted the spread of gender-insensitive and sexually suggestive content generated using AI tools.

Statistical analysis by BanglaFact revealed that 49% (35 items) of AI-generated content circulating online was politically themed. Among these, 68% focused on the Awami League, 17% on the BNP, 9% on Jamaat-e-Islami, and 6% on the NCP. Interestingly, most Awami League-related content portrayed the party in a positive light, seemingly aiming to enhance its public image.

The second most frequent targets of AI-generated disinformation were the Bangladesh Army and Police, featured in around 31% of the total content. Meanwhile, 17% of the materials referenced the interim government, and about 11% focused on the July mass uprising and its leaders.

AI-generated visuals also spiked during national crises or accidents. For instance, following the Uttara Milestone School and College plane crash, AI-manipulated content about the incident surged, accounting for roughly 10% of all samples collected during the research period.

Moreover, approximately 10% of the examined content involved gender-insensitive narratives, often targeting political figures or advisers to the interim government, with female politicians being disproportionately victimized by such malicious AI-driven disinformation.