NLPC presented a “Report on AI Data Usage Oversight” shareholder proposal today at the annual meeting of Meta Platforms, Inc., parent of Facebook and Instagram, that requested the Company to increase its reporting of the risks of “unethical or improper usage of external data in the development and training of its artificial intelligence offerings” and to disclose what safeguards the Company has in place.
The company’s board of directors opposed our proposal, as explained on page 67 of its proxy statement. NLPC responded to the board’s opposition statement in an exempt solicitation report circulated to the company’s shareholders.
Presenting the proposal at the meeting was Paul Chesser, director of NLPC’s Corporate Integrity Project. Audio of his presentation can be heard here, and a transcript of his three-minute remarks follows:
I’m Paul Chesser of National Legal and Policy Center.
Proposal Three asks Meta to do one thing.
To publish an annual report that tells us what management knows about the legal, financial, regulatory, and public-welfare risks of how Meta sources and uses external data to fuel its artificial intelligence products.
The company has guided 2026 capital expenditures of $115 billion dollars to $135 billion dollars, nearly double last year, with much of it pointed at AI infrastructure.
The Company’s AI strategy is built on the data inside its social-media empire — hundreds of billions of publicly shared images, tens of billions of public videos, and increasingly the relationships and context inside users’ personal lives.
Last month, Meta Superintelligence Labs released Muse Spark and marketed it as “personal superintelligence.”
So shareholders should ask a basic question.
What does the board know about the risks attached to that strategy, and what is it doing about them?
Meta’s track record on user data is not reassuring.
A record $1.2 billion euro fine for illegally transferring user data out of Europe.
A $725 million dollar settlement in the Cambridge Analytica litigation.
A privacy policy rewritten so that users’ images, videos, and text can be pulled into AI training without their explicit, opt-in consent.
And in March of this year, there were two jury verdicts in three days: $375 million dollars in New Mexico, plus a California verdict finding Meta and Google negligent in their design of products harmful to minors.
The board recommends voting against this Proposal, claiming existing privacy practices, disclosures, and committee oversight are sufficient.
But the board’s own opposition statement undermines that claim.
Meta admits that private chat content can be used to train its AI when “the user or someone in the chat chooses to share those messages.”
That is a structural consent loophole.
One participant’s choice sweeps every other participant’s words, photos, and context into Meta’s training pipelines — without their individual consent.
If Meta’s existing disclosures actually captured that risk, the board wouldn’t have to explain it away in the proxy statement.
But a proper risk report would.
Proposal Three does not ask Meta to slow down its AI investment.
It asks management to be candid with the people whose capital is funding it.
I urge shareholders to vote FOR Proposal Three.
Thank you.
All assertions made by Chesser in his above remarks are footnoted and can be found in NLPC’s exempt solicitation report circulated to Meta investors, here. An executive summary of the report can be viewed here.
Read NLPC’s shareholder proposal for the Meta annual meeting here.
Listen to Chesser’s presentation of the proposal at the meeting here.
