Meta Accused of Targeting Teens' Emotions in Ads

Testimony That Emotional States Were Used for Advertising

At a US Senate hearing on April 9, 2025, Sarah Wynn-Williams, Meta’s former Director of Global Public Policy, raised allegations concerning ads targeted at teenagers. She testified that Facebook had identified moments when users aged 13 to 17 felt helpless or self-critical, and passed that information along to advertisers.

She cited, as an example, a scenario in which a girl deleting a selfie was interpreted as a moment of insecurity about her appearance, prompting the platform to serve her a beauty ad. This differs from documented records confirming an actual case in which an ad appeared for a specific girl mere seconds after such an action. Meta denied the claims made in the testimony.

Looking at this allegation, I found myself thinking about the protection of adult users as well. I agree that teenagers deserve special protection, but I couldn’t help wondering whether the problem of exploiting anxiety or self-criticism for advertising can really be addressed by age alone.

The 2017 Report and the 2025 Testimony

In 2017, the Australian daily The Australian reported, based on internal Facebook documents, that the company had analyzed young users’ emotional states and presented the findings to advertisers. What sparked controversy was the claim that Facebook could pinpoint the exact moments when users needed a confidence boost.

Facebook responded at the time that it did not offer tools for targeting people based on emotional states. The company said the work was research meant to help marketers understand anonymized, aggregated data, and that it was never used for actual ad targeting. It did, however, acknowledge that the research had not gone through internal review procedures.

At a 2025 hearing, Wynn-Williams claimed there had been an even more direct use of the data. She alleged that Facebook shared teenagers’ vulnerable emotional states with advertisers and that anxieties about appearance and weight were exploited in ads. This is an allegation raised through an insider’s testimony — not a fact Meta has admitted to.

In September of that same year, Senator Grassley and other senators demanded further answers. Among the points of contention was research like the 2014 “Global Youth Study,” which analyzed young users’ emotions and behavior and was aimed at readers such as brand and sales teams. Confirming that such research exists is a separate task from confirming how its findings were actually applied to individual ad delivery.

Meta denied the 2025 testimony and reiterated its 2017 explanation. So we need to examine these questions separately: whether the company conducted emotional research, who received the findings, and whether they were actually applied to individual ad exposure.

Why Teenagers Need Special Protection

It’s only natural that criticism has emerged over allegations that advertisers exploited teenagers’ vulnerabilities. I want to be clear from the outset that teenagers do need special protection.

Young users are still developing the ability to recognize an ad’s persuasive intent and read it critically¹. In a 2004 report, the American Psychological Association cited this kind of difficulty in young children as grounds for recommending restrictions on advertising aimed at children under 8. But that research alone can’t support the claim that everyone’s ability to judge advertising critically is fully formed by age 18.

The United States’ COPPA is one example of a separate protection rule for children. In principle, it requires covered businesses—such as online services directed at children under 13—to obtain verifiable parental consent before collecting a child’s personal information. Since the age threshold this law covers differs from the 13-to-17 range raised at the hearing, a legal judgment tailored to the specifics of the allegations is needed.

There’s also the problem of receiving messages designed to trigger self-deprecation at a time when someone is especially sensitive about their appearance or social relationships. I think that if a platform identifies a user’s anxieties and turns them into a sales opportunity, that particular effect deserves separate scrutiny.

Strengthening protections for teenagers and safeguarding the rights of adult users aren’t mutually exclusive—both can be pursued at once. I just hope the latter doesn’t get dropped from the conversation.

Research on Users Beyond Teenagers

The hearing also surfaced an internal chat asking whether the company studied the emotional states of young mothers. A colleague confirms that it does and says they’ll look into whether there’s more research like it. This isn’t something we can dismiss as limited to research on teenagers alone.

That said, the phrase “young mothers” alone doesn’t tell us the age of every participant. And the mere record of such research doesn’t confirm that individualized ads were run based on their emotions. What I’m interested in is what emotions were inferred and where the results were used.

There’s also a case where Facebook experimented with conditions that influence how users express emotion. A study published in PNAS in 2014 adjusted the exposure to positive or negative posts in the News Feed of 689,003 users. Afterward, changes appeared in the ratio of positive to negative words in what users wrote.

This study showed that changing what’s displayed on screen can affect how users express themselves in writing. It wasn’t an advertising experiment that diagnosed emotional states or measured whether purchases increased. There was also controversy over whether consent for participating in the experiment was adequate.

Looking at these cases, I found myself thinking about the conditions under which inferring and intervening in emotion should be permitted. Even if stronger protections apply to teenagers, whether adults can understand and choose how their own data is used remains an important question.

The surveillance capitalism lens

Shoshana Zuboff, in her 2019 book The Age of Surveillance Capitalism, analyzed how companies convert users’ experiences and behavioral data into revenue. “Surveillance capitalism”² is a concept she’d been developing well before that.

Zuboff argues that companies collect behavioral data far beyond what’s needed to deliver a service, then use it to build products that predict human behavior. As competition intensifies, she argues, the incentive grows to move past prediction and actually intervene in behavior.

This lens helps clarify what we should be checking for in the mood-targeting allegations. If information was generated inferring that a user had lost confidence in herself, we need to look at who has access to that information and what it’s used for. But fitting a theory doesn’t prove any individual allegation.

Adults have emotionally difficult moments too. Assuming that, because someone is an adult, they can adequately evaluate and reject any ad, risks missing situations like this entirely.

Someone struggling after childbirth, or coping with job loss or having been a crime victim, should still be able to know how their information is used and make choices accordingly. That doesn’t mean we need to blanket-deny their judgment simply because they’re in such a situation. What I’m objecting to is a practice of concentrating sales messages on vulnerable moments, through inferences the person themselves would have no way of knowing about.

Oswarld’s Lens

From an advertiser’s perspective, showing a product when a customer feels the need for it matters. That’s exactly why Meta’s ad reach and conversion performance get such high marks. Personally, if I had to choose between spending my ad budget on Google Ads or Meta, I’d pick Meta. That’s my assessment of ad performance, not a claim that Meta is better for every campaign.

What I find more important, though, is the information gap between advertisers and users. Advertising that plays on emotion is nothing new—think of nostalgic Christmas ads or insurance ads urging you to prepare for risk. But when a platform infers a specific individual’s emotional state and times an ad to reach them without their knowledge, that’s a different scope of information use altogether.

I think we need to both verify whether the allegations raised at the hearing are true, and separately discuss what standards should govern the kind of data use that makes such advertising possible. Regardless of how Meta’s particular case is resolved, this is a question worth asking of every other platform too.

I agree with the case for protecting minors. As I’ve always said, turning twenty doesn’t suddenly make someone good at judgment in every situation. I believe adults, too, deserve to know how their own vulnerabilities are being inferred and exploited.

Wynn-Williams also testified that some Silicon Valley executives don’t let their own children use their companies’ products. That story unsettled me. Still, one executive’s personal choice can’t prove the entire allegation. What matters more is verifying how thoroughly companies actually explain the risks and their data practices to users.

Closing

Meta’s alleged emotional targeting needs to be examined by separating testimony, the company’s rebuttals, and the actual scope of the research. Alongside that, I want to discuss what disclosure and choice should look like when a company infers users’ emotions and puts them to commercial use.

Users should be able to find out what behaviors are used to infer what states, whether that information feeds into advertising, and whether they can opt out if they don’t want it. I think it’s possible to guarantee this right for adults while giving younger users special protections. So where’s your own line — how far would you be comfortable with companies inferring emotional states to target ads?

References & Further Reading

• Sarah Perez, “Meta whistleblower Sarah Wynn-Williams says company targeted ads at teens based on their ‘emotional state’”, TechCrunch, 2025.4.9. : A 2025 article covering the hearing testimony on emotional targeting and Meta’s response.
• Senate Judiciary Committee, “Grassley, Blackburn, Hawley Demand Answers on Meta’s Emotional Targeting of Children”, 2025.9.2. : Senate Judiciary Committee material disclosing additional evidence, including the 2014 “Global Youth Study.”
• Grassley to Meta, “Targeting-Teens” Letter, 2025.4.16. A letter containing primary-source material, including internal text messages between Meta employees.
• “Is Facebook Targeting Ads at Sad Teens?”, MIT Technology Review, 2017.5.1. : An article on the 2017 leaked Australian document — the origin point of this whole issue.
• Kramer, A., Guillory, J., & Hancock, J., “Experimental evidence of massive-scale emotional contagion through social networks”, PNAS, Vol.111 No.24, 2014. : The paper on the emotional contagion experiment involving roughly 690,000 people. It sparked controversy over whether participant consent was adequate.
• Shoshana Zuboff, The Age of Surveillance Capitalism, PublicAffairs, 2019. : Essential reading for understanding the structure of surveillance capitalism. It covers arguments about data collection and behavioral prediction and intervention. Also published in Korea under the translated title The Age of Surveillance Capitalism.
• Sarah Wynn-Williams, Careless People: A Cautionary Tale of Power, Greed and Lost Idealism, 2025. : The insider memoir that formed the basis for this hearing — the author’s firsthand account of internal decision-making at the company at the time.
• Sarah Wynn-Williams, written testimony submitted to the Senate, 2025.04.09. A 4-page opening statement focused mainly on the China business. Note: the filename contains “2024,” but this is not the full Q&A transcript.
• Meta, Comments on Research and Ad Targeting, 2017.04.30. The company’s official response regarding its emotional research and ad targeting.
• Cornell Chronicle, News feed: ‘Emotional contagion’ sweeps Facebook, 2014.06.10. The researchers explain the experiment’s design and measurements.
• APA, Task Force announcement on advertising to children, 2004.02.23.; FTC, COPPA guidance. Useful references on children’s understanding of advertising and the separate privacy rules that apply to them.
• Harvard Gazette, Interview with Shoshana Zuboff, 2019.03.04. The author explains the concept of surveillance capitalism in her own words.

The author is Oswarld (Kwangseob Ahn). Current roles: Adjunct Professor at Sejong University, Strategy Consultant at INLEVEL9. Career, research, books, and recent work are kept current on the About page. Latest · July 2026: HEMA-2: A Consolidation-Aware Tri-Memory Architecture with Multi-Channel Scheduling for Lifelong Conversational AI.

Footnotes

1. Ad literacy: the ability to understand what an advertisement is trying to persuade you of, and to critically evaluate the basis and intent behind its message. ↩

2. Surveillance capitalism: a concept Zuboff uses to describe an economic structure in which user experience is collected as behavioral data, which is then used to predict or influence behavior for profit. ↩