Prohibit pornographic deepfakes made without the person's consent
Artificial intelligence
Owners voting their values help. Others must carry most of it.
Why shareholder democracy is helpful
Making nonconsensual deepfake pornography a crime is the work of legislatures and courts. Platform conduct still shapes how far this material spreads, since companies decide what their services host and remove. Owners have a record on exactly this kind of harm. The Proxy Impact resolution asking Meta to report on combating online child sexual exploitation won about 56 percent of the non-management-controlled vote in 2021. ICCR's digital rights program engages platforms on harmful content and child safety online. No corporate blockade stands here. Congress passed the TAKE IT DOWN Act with broad bipartisan support, and the president signed it on May 19, 2025. The law makes publication of nonconsensual intimate images, including AI deepfakes, a federal crime. What remains is enforcement by prosecutors and the Federal Trade Commission, a government task.
How this was scoredHelpful
Helpful. Owners voting their values help. Others must carry most of it.
We have not yet written down the two answers behind this rating. Recording them on every item, so a reader can check the reasoning and not only the conclusion, is work in progress.
Every rating is the higher of two answers, one for what companies do and one for what they spend to stop the outcome, because the same vote reaches both. See the full method.
Sources
- Common Ground of the American People: Artificial Intelligence, Program for Public Consultation / Voice of the People, 2026 (study February 2024)
- Proxy Impact: Meta Shareholders to Vote on Resolution Asking Company to Combat Online Child Sexual Exploitation, PR Newswire (Proxy Impact), 2022-05
- Digital Rights and AI Accountability, Interfaith Center on Corporate Responsibility, accessed 2026-08-28
- President Trump Signs Take It Down Act Into Law, Latham & Watkins, 2025-05
What civil society organizations are helping owners on this
Has filed child safety proposals at Meta every year since 2020 on behalf of shareholder Lisette Cooper, asking the company to assess and report on child sexual exploitation risk and adopt child safety metrics.
ICCR members co-filed the Proxy Impact proposals asking Meta to report on how end-to-end encryption would affect its efforts to combat online child sexual exploitation.
Filed a 2024 proposal asking Meta to report on generative AI risks, citing sexualized nonconsensual AI-generated images circulating on Facebook and Instagram and ads for nudification tools on Instagram.
Exempt solicitation, Meta Platforms Proposal 6, Arjuna Capital via SEC EDGAR, 2024-04
4 campaigns on record
Ask the board's compensation committee to publish a report assessing the feasibility of integrating performance on improving child safety into Meta's senior executive compensation programme.
With Auris Gestion, on behalf of Auris Gravity US Equity Fund and Auris Diversified Beta, Congregation of St. Joseph, Daughters of Charity, Inc., Fonds Durocher / Durocher Fund, Everence Financial, on behalf of the Praxis Growth Index Fund, Northwest Women Religious Investment Trust, PFA Pension, the Hyde Cragmont 2015 Trust, the Linda C. Wisnewski Trust, Broz Family Investments LLC
Result: Proposal Ten at the annual meeting of 27 May 2026, not approved: 169,180,929 votes for, 4,776,963,503 against, 22,532,654 abstentions and 212,856,564 broker non-votes.
About 3.4 percent of the shares voted for and against. That is far below what the child safety reporting proposals drew at the same company, which ran from 16.3 percent in 2023 to 18.5 percent in 2024, and it is the first year the ask moved from disclosure to executive pay.
Ask the board to publish a report, at reasonable expense and excluding proprietary information, setting targets and quantitative metrics appropriate to assessing whether and how Meta has improved its performance globally on child safety impacts and actual harm reduction to children on its platforms.
With Durocher Fund, Dana Large Cap Equity Fund, Providence St. Joseph Health, Portico Benefit Services, Sisters of the Humility of Mary, PFA Pension, Benedictine Sisters of Cullman, Alabama, Sisters of St. Joseph of Peace, Everence Financial on behalf of Praxis Growth Index Fund, Missionary Oblates of Mary Immaculate, US Province, Sisters of the Presentation of the Blessed Virgin Mary of Aberdeen, South Dakota, The Daughters of Charity, Inc., The Congregation of St. Joseph, Hyde Cragmont 2015 Trust, Linda C. Wisnewski Trust
Result: Not approved at the 28 May 2025 annual meeting, as Proposal Nine: 669,263,526 votes for, 4,314,033,460 against, 10,306,035 abstentions and 204,772,865 broker non-votes, about 13.4 percent of the shares voted for and against.
Not approved at Meta's 2025 annual meeting.
Within one year, adopt targets and publish annually a report, at reasonable expense and excluding proprietary information, with quantitative metrics appropriate to assessing whether Meta has improved its performance globally on child safety impacts and actual harm reduction to children on its platforms.
With Durocher Fund, Providence St. Joseph Health, Lisette Cooper 2015 Trust, Linda C. Wisnewski Trust
Result: Not approved at the 29 May 2024 annual meeting: 925,650,012 votes for, 4,088,656,336 against, 10,232,329 abstentions and 204,808,280 broker non-votes, as certified in Meta's Item 5.07 Form 8-K. The lead filer puts that at 18.5 percent of the vote, and a majority of 59.1 percent of the vote not controlled by management, because Meta's dual class structure gives the chief executive ten votes a share.
Issue a report assessing the risks presented by the company's role in facilitating misinformation and disinformation disseminated or generated via generative AI, including sexualized nonconsensual AI-generated images.
With Open MIC, Eko
Result: About 17 percent of votes in favor; not passed.
Meta Investors Reject Scrutinizing Risk of AI Misinformation, Bloomberg Law, 2024-05
Related subjects on other maps
How this rating was made
Every item on every one of our maps is read against one question: how far does this depend on the people who own companies voting their values? We answer it twice, once for what companies do and once for what they spend to stop the outcome, and take whichever answer is higher, since the same vote reaches both. The score carries its reasoning and its sources so that a reader can check it rather than take it.
- Pivotal. Owners voting their values can deliver most of the outcome.
- Necessary. Owners remove an obstacle nothing else removes, and others must also act.
- Helpful. Owners voting their values help, and others carry most of it.
- Independent. This moves without owners. Other levers carry it.
The advocacy record on this page holds only what a source we opened says happened. Where we searched and found nothing, the page says so rather than leaving a silence. Where a campaign names a filer the source does not name, it says that too. Ratings are ours; the list of subjects is Common Ground of the American People's, and we link to their original.
How we score dependency, in full.