US shareholders have intensified their interest in the governance surrounding artificial intelligence (AI) through a rising number of targeted shareholder proposals.

Research shows shareholders proposals focused on AI are increasing. At the same time, US watchdogs have signed deals with key AI companies to inspect their products before they go to market. California is also set to push ahead with its own AI regulation.

FTI Consulting released figures this week showing the number of AI-focused shareholder proposals rose from seven in 2023 to 16 in 2024, signalling that concerns are growing.

Arnaud Cavé, a director with FTI, placed boards on notice that he expects interest to grow further following the first shareholder proposal in 2019, which addressed the potential for human rights risks in Amazon’s facial recognition software.

“With AI becoming ubiquitous in company operations, we expect more shareholder proposals and increased scrutiny from investors on AI practices in 2025,” Cavé says.

FTI says a “diverse” group of investors put together the shareholder proposals, with boards broadly recommending a vote against them.

None managed to achieve a majority, though a proposal at Netflix received the backing of 43.3% of shareholders and another at Apple found favour with 37.5%.

Board scrutiny of AI

The Netflix proposal asked the board to publish a report on the company’s use of AI in business operations and the board’s role in “overseeing AI usage”. The proposals sought publication of “any ethical guidelines the company has adopted” for its use of AI.

The board made a hard recommendation against the proposal, with its proxy statement saying the company was “committed to the responsible use of AI”. It added the proposal was “too vague” and could lead to publication of “competitively harmful information”.

Apple faced exactly the same resolution from the same source: AFL-CIO Equity Index Funds.

In the UK, a vote of 20% or more in favour of a shareholder proposal would require the board to engage with shareholders on the relevant issue. However, some in the City have called for the government to abolish this requirement.

AI shareholder proposals have yet to emerge in the UK or Europe but observers believe they could appear in the coming years. Cavé says: “Though restrictions on the ability to include proposals on the ballot vary market by market, what we know is that a number of European investors, such as Norges and LGIM, have publicly set out their expectations regarding AI governance and that AI has started to make its way into key topics of engagement between European companies and their investors.”

‘Milestone’ agreement

Regulation of AI is intensifying. Last week, the US AI Safety Institute signed an agreement with AI companies Anthropic and OpenAI to receive access to their new products before and after their release.

Elizabeth Kelly, director of the institute, said the agreement was a “milestone” in AI safety regulation, but “just the start”.

In California, new AI safety legislation has gone to governor Gavin Newsom for sign off. The rules will impose mandatory safety testing before a public release.

There has been growing concern about AI governance around the world. In the European Union, officials have been thrashing out new regulations for AI that will go to a vote of parliamentarians next year.

In the UK, the new government has already announced it will push ahead with legislation to beef up rules on AI. Some expect this to be a “statutory code”, requiring test data to be made publicly available.

There are many moving parts to the development of AI. Shareholder scrutiny is becoming one of them.