China’s next bogeyman: A deepfake video that makes its people think
If there is one thing that modern, autocratic China is very good at, it is responding to a perceived threat with all the resources it can muster. No half-measures.
Just ask the Uyghurs, China’s socially and technologically persecuted Muslim minority.
Now the nation wants to do what virtually every other country wants to do. The government wants to make it as close to impossible as it can to distribute unauthorized deepfakes and other fraudulent biometric content.
One authentic-looking video of autocrat-in-chief Xi Jinping frolicking nude through a meadow is arguably a bigger threat to the ruling Communist Party than Uyghurs.
So, it is not surprising that Beijing would begin a robust regulatory (and, later, social pressure) campaign to neuter deepfakes before they can darken public sentiment toward government leaders.
New rules, reportedly unique in the world, go into effect as early as March 1.
The effort ultimately could even lead to bureaucrats reading code to identify video, still-image and audio deepfakes. (Human parsing continues to be researched outside China, too.)
And that is just in China.
If Beijing successfully beats back deepfakes, governments of all stripes around the world will plunge ahead with similar campaigns to control fraudulent biometric development. At least some less-scrupulous governments likely would turn that control against dissidents, opposition politicians and rival nations.
China faces a herculean task, but few in the 1990s thought China could effectively wall off the global Internet to preserve the country’s hive mind.
The first step in the anti-deepfake program is here: a lengthy set of new regulations governing recommendation engines on content sites. Those are the personalized prompts that algorithms create to keep people on a site and consuming information.
Deepfakes can show up anywhere, of course, including in primary content, which leads to personalized prompts.
Targeting just recommendation engines could be a recognition on Beijing’s part that the battle against unauthorized deepfakes within its borders will be complex and maybe costly. It will need to be fought in pieces.
One of the regulators busy at this stage of the campaign is the Cyberspace Administration of China. Its translated, 35-article document can be found here.
Efforts in developed economies remains Balkanized and politicized, particularly in the United States, making a coherent defense against deepfakes more of an aspiration. Add vocal concern about preserving legitimate human rights, and a solution looks more distant still.
Coincidentally, a Swansea University law professor has been awarded a €1.5 million ($1.7 million) grant to see how the public’s perception of deepfakes impacts trust in crowdsourced evidence of human rights violations.
According to China’s new regulatory regime, it is platforms — not creators — that are held responsible for biometric trickery.
That strategy is the opposite of U.S. laws seeking to prevent falsified, libelous, criminal and other officially unwanted content from citizens, according to deep analysis of China’s playbook by non-profit news publisher Rest of World.
That puts the onus on, for example, ByteDance and its Weibo microblog service and Tautiao news aggregator, and Douyin, a TikTok-like service, according to Rest of World.