‘Existential Risks’: AI Anxiety Fueling Stream of Shareholder Proposals

The proposals will be about more transparency, what companies are doing with AI, and how boards are overseeing risks.

Public company shareholders this spring will vote on a bevy of investor proposals related to artificial intelligence (AI), a topic that rarely came up a year ago.

Investors already were deep into the process of crafting their 2023 proposals when AI burst into the public consciousness in November 2022 with OpenAI’s release of ChatGPT. But investors are ready his time around and are thrusting the topic, alongside other hot-button issues such as executive compensation and climate change, onto the agendas for shareholder meetings.

“Because this is such a new space, a lot of the proposals will be about more transparency and about what companies are doing with AI, and how boards are overseeing risks,” said Jamie Smith, the investor outreach and corporate governance director for the EY Americas Center for Board Matters.

The AFL-CIO, whose investment funds manage more than $12 billion, has submitted proposals seeking disclosure of AI standards to some of the nation’s largest and best-known companies, including Apple, Comcast, Disney, Netflix, and Warner Bros. Discovery.

Meanwhile, Trillium Asset Management, which specializes in socially responsible investing and manages $3.6 billion, has submitted a proposal to Google’s parent, Alphabet, that calls for the board’s Audit and Compliance Committee to take greater responsibility over AI product implementation and rollout.

While Alphabet in 2019 introduced AI principles that the company would abide by, “they have to take these principles, which are great on paper, and really bake them into the DNA of the company,” Trillium chief advocacy officer Jonas Kron told Treasury & Risk’s sister publication Law.com. “They have to make it part of the governance structure of the company.”

EY said in a report last month that the swift adoption of AI is posing a significant challenge for board audit committees. It said committee members “should inquire with management and internal audit regarding risk assessments around AI and related AI governance, including how risks around the ethical use of AI, accuracy of outputs, plagiarism, copyright, trademark violations, and protections of company IP were considered.”

Arjuna Capital, which specializes in sustainable investing and manages $319 million, partnered with the advocacy group Open MIC to submit proposals calling on Alphabet and Meta to produce in-depth reports on the dangers of generative AI’s deployment in misinformation campaigns and how the companies plan to address them.

“Alphabet and Meta need to assure billions of users and their shareholders that their management and boards are up to the task of responsibly managing [AI] technology,” Open MIC executive director Michael Connor said.

Arjuna’s Meta proposal asserts that while the company “has publicly acknowledged the risks of [generative] AI and outlined some guardrails, it continues to prioritize [generative] AI product development without addressing the existential risks posed by the technology.”

AI’s Broad Reach

Nicolas Dumont, a Cooley partner specializing in public companies, capital markets, and artificial intelligence, said the breadth of shareholder proposals reflects that AI defies easy categorization into the social, financial, or economic paradigms that investors have long targeted in their shareholder proposals.

“It could be anything regarding how the company intends to use AI, what kind of policies and procedures they have in place, whether they have ethics panels regarding the use of artificial intelligence,” Dumont said. “Shareholders have a lot on their mind.”

See also:

Many of the proposals that have come to light so far have been disclosed by the investors submitting them. The full scope of AI submissions will become clear in the coming weeks, as companies release proxy statements for their annual meetings.

Companies almost always recommend that shareholders vote against the proposals—arguing they already are making substantial progress on the issues raised or that the proposals veer into micromanaging the business—and they typically are soundly defeated. On top of that, the outcome of the votes is purely advisory.

Even so, investors say the filings are worthwhile because they throw a spotlight on issues they deem important and can pressure companies to rethink their practices.

‘Listen and Learn’

Last year’s proxy season did include a smattering of AI-related proposals. For example, one that Trillium submitted to Alphabet sought greater quantitative and qualitative disclosures of the AI and non-AI algorithms the business uses.

The proposal went down to defeat, garnering just 16 percent support. Alphabet had recommended shareholders vote “no,” arguing, “Our approach to algorithm disclosure requires us to focus on providing an appropriate level of transparency to mitigate related risks and bolster trust, balancing the critical business need to protect, first and foremost, our users and their privacy, and also the proprietary information that is foundational to our business—all of which could be misused in the wrong hands.”

But overall AI drew little attention, with Gibson, Dunn & Crutcher not even referencing the topic in the 30-page report it issued in July summarizing takeaways from the 2023 proxy season.

Even now, many investors are taking a “listen and learn” approach to AI, rather than using the shareholder-proposal process to pressure companies into taking specific actions, EY’s Smith said.

For instance, the proposals that the AFL-CIO submitted to media and tech giants merely asks for “AI transparency reports,” disclosing whether the company has adopted guidelines to protect workers, customers, and the general public from potential harms related to artificial intelligence.

Yet the AFL-CIO’s press release announcing the proposals makes clear that the labor organization is deeply worried about the implications of generative AI.

“The AI dehumanization of the American workforce threatens the very framework of the nation’s economy and endangers the existence of the already dwindling middle class while introducing the potential for discrimination in employment decisions,” the release says.

“Other perils include sudden mass layoffs due to job automation, the misuse of customer and employee private data, and the creation of ‘deepfake’ media content that may be used to disseminate false information.”

Apple and Disney argued to the SEC that they should not have to present the AFL-CIO proposal to shareholders, saying it covers “ordinary business operations” that are the purview of the board and management—one of the exceptions under the agency’s shareholder proposal rules. But the SEC disagreed, saying the proposal “transcends ordinary business matters and does not seek to micromanage the company.”

EY’s Smith said that whatever happens with this year’s AI shareholder proposals will help shape investors’ strategies for broaching the topic next year.

“We do expect them to be a significant part of this year’s shareholder proposal landscape, and we do think they will be closely watched in terms of how they fare and what kind of traction they get,” he said.

From: Corporate Counsel

Recommended Stories