ChinAI #234: The (Privacy) Cost of Being Fabulous?
China's viral AI-generated portrait app faces backlash on facial data collection
Greetings from a world where…
Does school really start this week?
…As always, the searchable archive of all past issues is here. Please please subscribe here to support ChinAI under a Guardian/Wikipedia-style tipping model (everyone gets the same content but those who can pay support access for all AND compensation for awesome ChinAI contributors).
Feature Translation: How Fabulous is Fabulous Duck (Miaoya Camera)
Context: If you’re interested in China’s data protection landscape, one must-follow public account is Southern Metropolis Daily’s Privacy Guard Team (隐私护卫队), which is connected to the Nandu Personal Information Protection Research Center. Recently, they covered the rise of Miaoya Xiangji (妙鸭相机), also translated as “Fabulous Duck,” an AI portrait generator that recently became one of China’s hottest apps. But: what happens with facial data after users upload 21 photos of their face to create an exclusive digital avatar?
***Thanks to Ben Jiang, a reporter with the Tech Desk at the South China Morning Post, for contributing this week’s translation and analysis (lightly edited by me). Check out his reporting here.
Key Takeaways from Ben:
When it comes to artificial intelligence regulation, the Chinese leadership has taken a very pragmatic approach, that is, their primary focus would be to harness the prowess of the technology towards upscaling its industries and businesses to keep the growth engine humming, while at the same time to root out any side effects they don’t want or at least keep those at bay.
In the case of generative AI, the tight lid was more on content safety, which includes items from the generative AI regulation that came into effect last week that mandate AI-generated content to “hold onto core socialist values” and not to produce content that subverts the CCP-held regime, damage the country’s image, and undermine national unity or social stability.
While there are also rules related to seeking consent for personal information from rights holders and conforming to laws and regulations to legally source training data and materials, there was no explicit definition for the breaches. At the same time, such AI services providers are held accountable for the content they generate, to immediately suspend their generative services, eliminate the content, and to report to the authorities.
Key Passages I want to highlight:
This case has echoes of another viral sensation from 2019: Chinese face-swapping app Zao. That case led to significant backlash over lack of data privacy protections. Miaoya’s terms of use contained similar clauses that would give the app permanent, irreversible rights to content upload and generated by users, including to use as training data.
Miaoya’s response to the backlash: “Miaoya announced on July 20 on its Weibo account that it has received feedback over its term of use, and that the original contract terms contained incorrect information. It has made amendments immediately, and ‘promises earnestly to everyone that, your uploaded photo will only be used to create digital-self, it will not be extracted or used for recognition or other purpose. Once a digital-self is created the photos will also be deleted.’ “
Interesting note that connects this Miaoya case with the China’s algorithm registry from Ma Ce, a partner at Kin Ding Law Firm: “According to Ma, China currently has policies for companies to file their generative AI algorithms with regulators, and for them to make clear the risks of how their algorithms could be abused, which involves the right for a user to review, copy, and delete information. This is what Miaoya will need to clarify in its user terms and algorithm registry filings.”
FULL TRANSLATION: Behind the Popularity of Miaoya Camera (Fabulous Duck): Experts say that AI portrait technology is not new, and data concerns need to be taken seriously
ChinAI Links (Four to Forward)
Must-read: The All-American Myth of the TikTok Spy
For Wired, Yangyang Cheng’s latest essay is essential reading:
During the Mao era, Western writers used the phrase “blue ants” to describe the Chinese population, in reference to the light navy uniforms of the time. This image of the Chinese people as faceless, mindless automatons has also shaped American perceptions of Chinese spycraft, most notably in the so-called “thousand grains of sand” theory. Proposed by FBI analysts in the 1980s, the metaphor goes like this: To gather intelligence about a beach, the Russians would send in a submarine and frogmen, the Americans would use satellites, and the Chinese would simply dispatch a thousand tourists, each collecting a single grain of sand. The theory has been refuted—Beijing’s intelligence operations are not so different from those in other countries and mostly rely on professionals—but not before it had been hailed as doctrine for decades and cast a distorting light on every Chinese visitor as a potential foreign agent.
This February, during the first hearing by the newly established House Select Committee on China, former national security adviser H. R. McMaster described the pervasiveness of Chinese industrial espionage in graphic terms: “If you bolt your front door, they’re coming through the window. If you bolt your windows and put up screens, they’re going to tunnel under your house.” According to the retired general, the “many vectors of attack” demanded “a holistic approach.” As I listened to the hearing, I wondered how many who had heard these words pictured an army of ants creeping through the cracks, sweeping up every grain of sand.
Should-read: New Chinese Facial Recognition Regulations Could Shield Citizens From Surveillance Capitalism
For Forbes, Johanna Costigan has a good readout of China’s draft regulations on facial recognition technology: “They enable continued state surveillance—and overt government exceptionalism. But they also grant individuals the right to protect their personal data (in this case, the especially intimate data of one’s face) from businesses that stand to market and profit off people’s likeness..While a half completed task shouldn’t be praised, it is a half step further than the United States has taken.”
Should-read: A puzzle stumps statisticians: How much does China actually spend on R&D?
I stumbled upon this scibus article by accident — I was translating some old articles that Changlin Gao had published for an ongoing research project and wanted to get his bio (apparently, he’s now the minister counselor for science and technology at the Mission of China to the EU).
Anyways, this article covers a quiet notice issued by the OECD, which maintains an influential database of science and technology indicators, that “it is suppressing publication of annual Chinese R&D statistics dating back to 2019” due to possible inconsistencies with different measures of R&D spending and researchers in the higher education sector. Richard Hudson gives the broader context:
But why the fuss, amplified on social media in the small global community that follows these kind of statistics? In part, it’s because the OECD rarely halts publication of so many statistics from so important a country as China. And in part, it’s because China’s rising R&D investment – however big it is – is anxiously watched by tech policy experts in Washington, Brussels and other Western capitals.
Should-read: CSET Data Brief - U.S. and Chinese Military AI Purchases
Abstract: This data brief uses procurement records published by the U.S. Department of Defense and China’s People’s Liberation Army between April and November of 2020 to assess, and, where appropriate, compare what each military is buying when it comes to artificial intelligence. We find that the two militaries are prioritizing similar application areas, especially intelligent and autonomous vehicles and AI applications for intelligence, surveillance and reconnaissance.
By: Margarita Konaev, Ryan Fedasiuk, Jack Corrigan, Ellen Lu, Alex Stephenson, Helen Toner, and Rebecca Gelles.
Thank you for reading and engaging.
These are Jeff Ding's (sometimes) weekly translations of Chinese-language musings on AI and related topics. Jeff is an Assistant Professor of Political Science at George Washington University.
Check out the archive of all past issues here & please subscribe here to support ChinAI under a Guardian/Wikipedia-style tipping model (everyone gets the same content but those who can pay for a subscription will support access for all).
Any suggestions or feedback? Let me know at chinainewsletter@gmail.com or on Twitter at @jjding99