AI Serving Dictators: Freedom House Report

For thirteen consecutive years, global internet freedom has been waning. Iranians are experiencing intensified restrictions, with internet services curtailed, platforms like WhatsApp and Instagram blocked, and increased surveillance to suppress anti-government sentiments. China remains at the forefront of digital oppression, leaving little to no online freedom for its citizens. Myanmar shadows China in these repressive measures. Meanwhile, the Philippines is witnessing a decline in internet freedom, with President Rodrigo Duterte using the anti-terrorism act to clamp down on opposition news outlets. Even in Costa Rica, previously celebrated for its online liberty, politicians resort to employing online trolls to target and discredit independent media.

Over the last decade, the count of governments resorting to chatbots for influencing online discussions has seen a twofold increase, now totaling 47. Unsurprisingly, AI, proficient in generating text, audio, and images, is at the forefront of this movement. As AI tools become increasingly sophisticated, more affordable, and widely available, their implications are vast. According to Freedom House, this nascent technology was leveraged in 16 nations last year alone to heighten civic unease, discredit adversaries, and shape public conversation.

In the backdrop of tensions between pro-government factions and the opposition in Pakistan, former Prime Minister Imran Khan offered his take on the unfolding events.

“The way Pakistani women stood up for Haqeeqi Azadi, they will be remembered and become part of our democratic history. Also what will never be forgotten is the brutality of our security forces and the shameless way they went out of their way to abuse, hurt and humiliate our women. Hundreds are languishing in jail in terrible conditions. This too won’t be forgotten,” wrote Khan.

Moreover, Khan displayed a video where his name echoed in the chants of supporters, while an unidentified woman fearlessly stood up to the police. But the twist came when it was later discovered that the portrayal of the valiant Pakistani lady had been synthetically crafted using AI.

Even in democratic nations, the use of deepfake technology is not uncommon. U.S. President Donald Trump and Florida Governor Ron DeSantis have employed deepfake videos and audio to smear each other's reputation. Trump set up a fabricated event in Twitter Spaces, where one could purportedly hear a conversation involving voice impersonations of DeSantis, Elon Musk, Dick Cheney, and even outlandish figures like Satan and Hitler (!).

DeSantis, on the other hand, circulated fake evidence suggesting a bond between Trump and the face of pandemic restrictions, Dr. Anthony Fauci.

Yet, there's another facet to this issue. Experts in the AI and disinformation fields highlight that harm can arise not just from deepfakes themselves, but also from baseless accusations of their use. A case in point is Gabon, West Africa. In 2018, the health of Gabon's president, Ali Bongo Ondimba, deteriorated, and he withdrew from public appearances. Rumors began circulating in Gabon that the leader had passed away. People anxiously awaited Ondimba's New Year message, and when he finally addressed the public, his appearance seemed slightly off. Some media outlets conjectured he might have had a stroke. However, opposition voices amplified the idea that the video was a deepfake. Consequently, the nation teetered on the brink of a military coup, which fortunately was swiftly suppressed. Such political turmoil was incited by unfounded allegations asserting the ruler is an imposter.

In addition to the threats posed by deepfakes and bots, AI has been increasingly leveraged by governments to bolster online censorship. This phenomenon is most notable in technologically advanced authoritarian regimes. Currently, there are 21 countries that either mandate or encourage digital platforms to utilize machine learning for the removal of undesirable content.

As of now, there are no established laws specifically governing AI. Stakeholders in the industry are determining their own guidelines based on their discretion. This kind of “self-regulation” has rendered human rights increasingly vulnerable. To ensure a free internet remains, democratic leaders, together with members of the civic society, must put forth rigorous standards for both governmental and private developers and AI users. This is a sentiment strongly endorsed by Freedom House.