China to impose ‘ethics reviews’ on research covering humans, animals and AI
- Seven types of research will require verification from third-party experts, including ‘algorithmic models … with the ability to shape public opinion’
- Research involving data and algorithms, two critical components in large language model development for AI, are also subject to the review process
China will require its universities, hospitals, enterprises and research institutions to conduct “ethics reviews” of any research covering humans, animals and artificial intelligence (AI), according to regulations published by the central government this week.
The new rules, effective December 1, list seven types of ethics research that will require additional verification from third-party experts, including “algorithmic models, applications and systems with the ability to shape public opinion and guide social awareness” and “clinical research on invasive brain-computer interfaces”, according to a notice published by the Ministry of Science and Technology.
Other research requiring additional oversight from regulators includes “synthesis of new species” that may have “significant impact on human health, values and the environment”, human stem cells in animal embryos, genetic editing, human-computer interactive systems if they impact human mental and physical health, as well as “automated decision-making systems with security and health implications”, according to the provisional regulations which were also endorsed by the Ministry of Education, the Ministry of Industry and Information Technology, and the science committee of the Central Military Commission.
It is China’s first set of regulations covering how institutions should conduct “ethics reviews” of their research, in line with Beijing’s efforts to set guardrails for scientific research. In one high-profile case, Chinese scientist He Jiankui sparked global controversy when he created the world’s first gene-edited babies. He was sentenced to three years’ imprisonment in 2019 for “illegal medical practices”.
The regulations are aimed at promoting science and technology “for good”, the science ministry said in a separate statement. The rules are intended “to standardise ethics reviews in scientific research … to control ethical risks, and to promote responsible innovation,” according to the guidelines.
Under the rules, every institution involved in “sensitive research areas” will have to set up an ethics review committee with a minimum of seven members. The committee will have the power to cancel any research project submitted for review.
Research involving data and algorithms, two critical components in large language model development for AI, are also subject to the review process, according to the regulations. China’s ministry of agriculture and the health ministry, as well as the Chinese Academy of Sciences, also signed off on the regulations.
The design of algorithms and AI models should adhere to fair, transparent, reliable and controllable principles that are in compliance with relevant state requirements, echoing similar requirements from the cyberspace administration, China’s powerful internet watchdog.
China is ahead of the rest of the world in designing regulation to oversee the development and use of generative AI technologies. Internet regulator the Cyberspace Administration of China, for example, has introduced a de facto licensing regime for AI chatbots to operate in the country.
Other nations are still debating how to regulate the promising, but also potentially disruptive technology of generative AI.
