Yesterday, ISS Governance published results from its annual global benchmark policy survey, which is part of the process for developing its annual voting policy updates. Among other things, the survey asked investors and companies (non-investors) to share opinions on AI oversight and related disclosures. Here are a few of the key takeaways on investors’ views:

Respondents were asked if expecting a company significantly using AI to use a global framework (for example, OECD AI Principles, NIST AI RMF, etc.) for assessing AI-related risks is appropriate at this time. Non-investor respondents overwhelmingly selected the option “It is probably premature for most companies.” ( 84 percent) while only 16 percent opted for the remaining alternative “It is probably timely for most companies.”

On the other hand, 58 percent of the investor respondents supported the “It is probably timely for most companies” option, and the remaining 41 percent the “It is probably premature for most companies” option.

Respondents’ perspectives were sought on whether companies should publicly share how their boards are overseeing AI business or AI implementation systems with the goal of managing AI-related risks. The preferred option for both categories (54 percent for the investor respondents and 73 percent for the non-investor respondents) was “Only in cases where AI plays a significant role in the business or business strategy (where businesses already have or plan to implement significant AI use).” The answer “In all or most cases – companies/boards which do not consider it relevant can disclose and explain their rationale.” was chosen by 43 percent of investors and 13 percent of non-investors.

Respondents were also asked to what extent a board’s public disclosure of its AI oversight measures indicates its depth of understanding of AI-related issues and risks. Both investor and non-investor respondents expressed a preference for the option “Public disclosure alone does not necessarily imply a board’s solid understanding of AI.” ( 69 and 53 percent, respectively). Slightly less than one-third of non-investor respondents (29 percent) opted for the alternative “There is little general correlation between disclosure and understanding.”

The survey also asked about AI-related board expertise:

40 percent of investor respondents and 25 percent of non-investor respondent opted for the answer “Only companies where AI is central to their core business or poses significant risks would need an AI expert or dedicated committee.“, while the option “Unless AI is central to their core business or poses significant risks, it is sufficient for most boards to have access to external AI advisors when needed.“, was the favored choice of 38 percent of the investor respondents and of 58 percent of the non-investor respondents.

On The Proxy Season Blog for our members, Meredith recently shared a disclosure resource for AI-related opportunities, risks, and governance – and we’ve been tracking trends in disclosures – as well as AI-related shareholder proposals.

We’re also continuing to post resources on compliance & disclosure issues in our “Artificial Intelligence” Practice Area on this site. And don’t forget to sign up for our free blog – The AI Counsel – for daily updates on evolving AI & emerging technology risks.

– Liz Dunshee