YouTube’s algorithmic suppression of Iranian digital narratives reflects geopolitical censorship in AI-driven content moderation

Indigenous digital sovereignty frameworks, such as those developed by Māori and First Nations communities, emphasize collective ownership of cultural narratives and reject platform-based censorship as a form of neo-colonial erasure. The Iranian case echoes global patterns where Indigenous and non-Western digital expressions are flagged as 'extremist' or 'misleading' due to algorithmic bias trained on Western datasets. Traditional knowledge systems often prioritize context and relationality over binary 'truth' claims, making them incompatible with Silicon Valley’s reductive moderation models.

The YouTube ban reflects a century-long pattern of Western powers suppressing non-Western cultural and political expression, from colonial-era book burnings to Cold War-era radio jamming. Iran’s digital censorship struggles mirror its 1953 coup (orchestrated by Western intelligence) and the subsequent imposition of sanctions that severed cultural and academic exchanges. The use of AI to enforce these bans is a modern iteration of the 19th-century 'civilizing mission,' where technology is deployed to police the boundaries of acceptable discourse.

In China, platforms like Weibo and Douyin deploy AI moderation to suppress dissent while promoting state-aligned narratives, creating a mirror image of YouTube’s approach. Nigerian digital activists use encrypted messaging apps to circumvent government surveillance, framing digital resistance as a form of *ubuntu*-inspired collective care. The Latin American concept of *buen vivir* (living well) critiques the commodification of digital spaces, advocating for platforms designed to serve communal rather than corporate interests.

Research on algorithmic bias shows that content moderation systems trained on Western datasets disproportionately flag non-Western languages, religious symbols, and political discourse as 'violent' or 'hateful.' Studies on digital repression indicate that sanctions regimes (like those targeting Iran) correlate with increased self-censorship among online communities, even when no formal ban exists. The Lego-style AI videos, while visually innocuous, may trigger moderation due to their association with state-aligned narratives in training data, highlighting the 'guilt by association' problem in AI classification.

The Lego-style AI videos can be read as a form of *ta'zieh* (Shia passion plays), where digital puppetry becomes a medium for communal storytelling under censorship. In Sufi traditions, art is a tool for transcending oppressive structures, a concept mirrored in Iranian digital artists’ use of absurdity and play to subvert algorithmic surveillance. The ban itself evokes the historical destruction of cultural artifacts (e.g., the Buddhas of Bamiyan), framing digital erasure as a contemporary form of iconoclasm.

If unchecked, AI-driven content moderation will entrench a bifurcated internet: one for Western-aligned narratives and another for the rest, with 'neutral' platforms acting as de facto censors for U.S. foreign policy. Scenario modeling suggests that non-Western platforms (e.g., Iran’s Aparat, Russia’s VK) will develop parallel moderation systems, fragmenting global digital commons into geopolitical blocs. The rise of decentralized social media (e.g., blockchain-based platforms) could offer alternatives, but risks replicating existing power imbalances without radical redesign.

Iranian feminists, LGBTQ+ creators, and ethnic minorities face double censorship: from state authorities and platform algorithms that conflate dissent with 'terrorism.' Iranian Kurds and Baloch communities report higher rates of account suspensions, reflecting how digital repression targets already marginalized groups. The focus on state-aligned content obscures the voices of Iranian dissidents who use AI tools to document human rights abuses, further silencing those most at risk.