Last year was nothing short of a feast for the capital market of the artificial intelligence (AI) industry, with a string of leading AI firms making their debut.
Cloudwalk, MEGVII, Deep Glint, Intellif and Orbbec all successfully registered for listing on the Star Market. By the end of the year, SenseTime, one of the “four little dragons of AI”, led the race to successfully complete listing on the Stock Exchange of Hong Kong, setting a new trend for the development of the AI capital market.
Policies in full swing
Regulatory policies for AI were also in full swing in 2021. On 5 January, the National Information Security Standardisation Technical Committee issued the Practice Guide to Cybersecurity Standards ‒ Guidelines on the Prevention of Ethical Security Risks from Artificial Intelligence.
On 25 September, the National Governance Committee for the New Generation Artificial Intelligence issued its Code of Ethics for New Generation of Artificial Intelligence, aiming to integrate ethics into the whole life cycle of AI with ethical guidance for natural persons, legal persons and other organisations engaged in AI-related activities.
On 17 December, the Central Committee for Comprehensive Deepening Reform adopted Guiding Opinions on Strengthening Ethical Governance of Science and Technology (the “Guiding Opinions”), which prioritises ethics, agile governance, feasibility and open co-operation as basic requirements of scientific and technological activities, to effectively prevent possible ethical risks caused by scientific and technological innovation – stressing that technology should be used for goodness.
In the same year, the Data Security Law and Personal Information Protection Law were promulgated and implemented, both also closely related to data compliance and scientific and technological ethics required by regulatory authorities for the AI industry.
In the above-mentioned context, examination and review of IPO registration followed suit. MEGVII was the first to face inquiry about the ethics of science and technology during its registration on the Shanghai Stock Exchange (SSE) in April 2021.
In a section headed “data compliance and scientific ethics” in the SSE’s inquiry letter, one of the questions required the company to disclose “organisational structure, core principle, internal control and implementation of the ethics of AI”, and explain “measures of the company to implement relevant responsibilities and comply with ethics-related norms and standards and the status of their implementation in R&D and business development”.
It further instructed the company to analyse its measures and plans to ensure its AI technology was controllable and in line with ethical norms, as well as any ethical risks it faced.
From June to July 2021, the SSE successively required explanation from AI companies like Cloudwalk, Deep Glint and Orbbec about “the measures and plans of the company to ensure that the AI technology is controllable and in line with ethical norms”, asking some to disclose the ethical risks.
Just one year ago, when Yitu Technology, another of the “four little dragons of AI”, was listed there was no such inquiry about scientific and technological ethics.
SenseTime, which took the lead in completing listing on the HKEX, disclosed in its prospectus submitted in August 2021 that the company adhered to “the principles of sustainable development, people-oriented and technology controllability” as the ethical principles of AI – disclosing its “ethical measures on AI” and assured compliance with ethical norms of science and technology by way of organisational structure and governance, internal measures and co-operation with third-party institutions and think tanks.
Key regulator concerns
Based on the above-mentioned comparative analysis, the author summarises the key points of regulators’ concern in IPO reviews of AI companies as follows:
(1) Controllability of AI technology. Similar to how “out-of-control risk” is regarded as a primary risk for ethical security of AI in the above-mentioned guidelines, “ensuring controllability and trustworthiness” is listed in the above-mentioned code of ethics as one of the basic ethical norms that all AI activities should follow to ensure that AI is at all times under human control. In this respect, regulators focus on how prospective listed companies achieve “controllability” in terms of system, technology and internal control mechanism.
(2) R&D and business development. Existing ethical norms mainly include the above-mentioned guidelines and guiding opinions, which have not entered legislation. As the guiding opinions point out regarding “perfecting the governance system of science and technology ethics”, in future we should “strengthen the legislative research in key areas of science and technology ethics, and upgrade important scientific and technological ethical norms into laws and regulations in a timely manner”.
Therefore, at present, concerns of regulators focus on: Whether a review mechanism of science and technology ethics has been established within the company; whether this mechanism can be effectively implemented; and whether awareness of science and technology ethics can be effectively raised among employees.
(3) Ethical risks for companies. The guidelines state that ethical security risks of AI mainly include out-of-control risks, social risks, infringement risks, discriminatory risks and liability risks. Judging from the publicly available prospectuses of the above-mentioned AI companies, discussions mostly centre on liability risks and social risks that may be brought about by technology abuse (misuse), while few words were spared for other risks.
Expect more stringent requirements
It is worth noting that the code of ethics was promulgated at the end of September 2021, and was not mentioned in earlier inquiries, replies and prospectuses of those IPO companies.
If referring to the code of ethics in future, the SSE may impose more stringent review requirements, as the code clearly states that ethical norms are to be observed in specific activities of AI including “management norms, R&D norms, supply norms and use norms” – not just ethical norms and standards to comply with in the above-mentioned process of “R&D and business development”.
Accordingly, when designing measures to ensure compliance with ethical norms, companies should not be content with establishing an internal ethical review mechanism, staff training and management, but should also include external review and management on suppliers, clients and partners, so as to meet standards provided in the supply and use specifications.
Huang Hui is an associate at East & Concord Partners
East & Concord Partners
Suite 29/F Hongcheng International Plaza, 358-396 Juyuan Road, Hangzhou, 310020, P. R. China, P.R.China
Tel: +86571 8501 7011
Fax: +86571 5801 7085