In a move that mirrors a debate in the U.S., China‘s communist regime is taking steps to regulate the exploding field of generative artificial intelligence, announcing plans to enforce rules to label and restrict information created by powerful new tech tools.
The Cyberspace Administration of China said Friday the new enforcement push will include clear indications marking AI-generated content, policing internet news and information produced without approval, and a fresh crackdown on information the regulator determines to be false.
The Chinese regulator’s “special actions” will focus on specific areas of the internet and research, and include heightened scrutiny for internet “trolls” and rumors that the communist regime dislikes, according to an English-language translation of the regulatory notice.
The notice reveals that despite the vast ideological differences and sharp geopolitical competition, the Chinese government shares many of the concerns that Biden administration officials and lawmakers on Capitol Hill have voiced about new generative AI tools involving the spread of misinformation and deceptive content.
Both countries, as well as regulators for the European Union, are struggling to find the balance between encouraging AI’s positive aspects and curbing abuses and problems that have already surfaced.
Generative AI tools include popular products such as ChatGPT, which use powerful models to create text, images and videos in response to queries from users, often with startlingly human qualities.
Some of China’s planned rules resemble proposals from American lawmakers, such as China’s push to mark AI-generated content.
Sens. Brian Schatz, Hawaii Democrat, and John Kennedy, Louisiana Republican, introduced the AI Labeling Act last summer to require AI warning labels, citing what they said was a desire to stop scammers from duping Americans.
While the U.S. lawmakers’ bill stalled and other proposals remain under debate, Beijing has continued to formulate aggressive regulations. The state-affiliated Global Times said the new push announced Friday builds upon interim regulations governing generative AI dating back to August 2023, part of the country’s first comprehensive AI-focused regulation.
International perceptions that China is cracking down on AI may mistake the communist regime’s bark for its bite, according to University of Hong Kong law professor Angela Huyue Zhang.
In a research paper published last month, Ms. Zhang said China‘s fear of losing the AI race to American and Western rivals, coupled with the escalating chip embargo on Chinese AI firms spearheaded by Washington, are diminishing the communist government’s enthusiasm to restrain domestic Chinese players and actually enforce the proposed AI rules.
“As evidenced by its permissive stance over the abusive use of facial recognition technology, Chinese regulators have favored a light-touch approach to AI regulation in practice,” Ms. Zhang wrote. “Similarly, Chinese courts are trying to prop up the AI industry.”
Ms. Zhang said China’s AI development rivals the U.S. in certain commercial and national security AI applications, for things such as facial recognition, voice recognition and AI-enhanced drones.
But Ms. Zhang said the U.S. far surpasses China in other AI domains, including those leveraging “large language” models, which are the powerful algorithms making generative AI tools possible. She said industry analysts estimate China’s AI companies are behind their U.S. counterparts by two or three years.
“China’s strategic lenient approach to regulation may, therefore, offer its AI firms a short-term competitive advantage over their European and U.S. counterparts,” Ms. Zhang wrote. “However, this leniency risks creating potential regulatory lags that could escalate into AI-induced accidents and even disasters.”
The fear of losing a competitive advantage by imposing too strict rules on AI has surfaced in the U.S. as well.
Small AI startups have complained that President Biden’s October executive order will hurt their businesses before they can get off the ground, according to Esube Bekele, an official with In-Q-Tel, the Virginia-based nonprofit venture capital company that invests in companies to help keep U.S. intelligence agencies at the cutting edge of information technology
Concerns about Mr. Biden’s AI rules are also mounting on Capitol Hill. The House Committee on Oversight and Accountability has scheduled a hearing later this month on what it is billing as “White House Overreach on AI.”