Shenzhen pilots regulation to classify and regulate AI technology according to r...

Shenzhen pilots regulation to classify and regulate AI technology according to risk levels- PingWest

Shenzhen pilots regulation to classify and regulate AI technology according to risk levels

4 hours ago

Shenzhen is piloting a new regulation aimed at regulating different artificial intelligence (AI) technologies according to their risk level classifications. Similar regulation already exists in Europe. This Shenzhen regulation, should it go into effect, would be a first in China.

Details: detail of the proposed regulation with regard to the classification method is yet to be announced, as texts of a latest draft version are largely a call to action rather than a elaborate handbook. 

However, the draft version does mention that high risk AI application will be thoroughly evaluated before adoption and operated on continuously running risk models, while the adoption of medium-to-low risk applications should accelerated and regualted on a record-keeping and disclosure system.

Other interesting articles from the proposed regulation also mention that Shenzhen is looking to invite more foreign AI talents, and further expand the market economy of the AI industry by allowing more advanced technology suppliers from overseas to participate in government projects.

The regulation also calls for improved explainability of AI algorithms, especially when the commercial adoption of them affects public interest.

Context: the proposed Shenzhen regulation follows the suit of a similar one submitted to the European Commission, called Regulation Laying Down Harmonized Rules on Artificial Intelligence, which would establish rules on the development, placing on the market, and use of AI systems across the EU.

Although the EU regulation said specifically that public ahothorities should not use AI systems for "social scoring", which the Shenzhen regulation did not mention, the two do agree with each other on one thing: classifying and regulation AI technologies according to their risk levels. Both mention that high-risk AI technologies should promise explainability and be subject to stricter governance practices, such as risk assessment, to ensure safety.

SHARE