Misinformation advertising: how does ad money reach false news?
Part of Brand suitability
The short answer
Ads can reach misinformation sites through automated buying, which spreads a campaign across sites nobody reviews. A 2024 study in Nature found ads on 74.5 percent of the misinformation sites it examined. Brands reduce the risk by defining misinformation narrowly, as verifiably false claims tied to real harm, and by buying from reviewed site lists rather than the open web.
Key takeaways
- In a 2024 Nature study, 2.7 percent of US misinformation sites rated by one source had a paywall, against 25.0 percent of other news sites.
- A 2024 Nature study found companies using digital ad platforms were about ten times more likely to appear on misinformation sites than companies that did not.
- Only 20 percent of decision makers surveyed believed their company's ads had recently appeared on such sites, and consumers in an experiment moved away from brands that did.
- A workable policy targets verifiably false claims linked to harm, not opinions, contested policy or reporting about falsehoods.
- Who rates a site, and on what evidence, matters as much as the definition.
How does advertising end up funding misinformation?
Programmatic ads on the open web are bought by software, not by people choosing sites. A buyer sets targeting and a budget, and the buying platform bids on matching impressions across the sites its exchanges offer. A site publishing false health cures or invented election news can sign up with an ad network, add ad code and start earning from the same campaigns that run on national newspapers. Nobody at the brand chose the site, and nobody at the agency may have heard of it.
The business model explains why this matters. Among the sites studied by Ahmad and colleagues in Nature (opens in a new tab) (June 2024), 2.7 percent of US misinformation sites rated by one of the study's sources had a paywall, against 25.0 percent of other news sites. Ad revenue is a lever that reaches the supply of misinformation rather than its readers.
What have studies found about ads on misinformation sites?
A detailed recent study is the Nature paper by Wajeeha Ahmad, Ananya Sen, Charles Eesley and Erik Brynjolfsson. It combined advertising data from 2019 to 2021 with lists of misinformation domains from two independent rating organizations and earlier research, covering 5,485 websites, 1,276 of them classed as misinformation, and 42,595 advertisers. It then ran an experiment with consumers and a survey of company decision makers.
| Question | What the study found |
|---|---|
| How many misinformation sites carried ads? | 74.5 percent of misinformation sites in the sample were monetized by advertising from 2019 to 2021 |
| How widespread was advertising on them? | Companies advertising on misinformation sites made up 46 to 82 percent of the companies in each industry studied |
| What role did ad platforms play? | Among the 100 most active advertisers, 79.8 percent of those using digital ad platforms in a week appeared on misinformation sites that week, against 7.74 percent of those not using them; about ten times more likely |
| Did decision makers know? | Only 20 percent believed their own company's ads had recently appeared on misinformation sites; about 81 percent of those who asked for a check had in fact appeared there |
| Did consumers react? | In an experiment, consumers told about a company's ads on misinformation sites moved away from it; those who switched gave up about 39.4 percent of a US$25 gift card's mean value |
The authors also found that decision makers strongly preferred to avoid misinformation sites once informed, and that telling them how platforms place ads raised demand for a platform based fix. The practical reading for a buyer is that the exposure in the study was largely unchosen. It comes from buying methods, so it can be reduced by changing how sites are selected.
How should a brand define misinformation in its policy?
Definition
- Misinformation (brand safety)
- In the legacy suitability framework's words, the presence of verifiably false or willfully misleading content that is directly connected to user or societal harm. Both parts matter: the claim must be checkably false, and it must be linked to harm.
That definition, from the Brand Safety Floor and Suitability Framework built by GARM before the World Federation of Advertisers discontinued it in August 2024, is the one this page uses. The Brand Safety Institute's copy (opens in a new tab) grades the category like any other: glamorization or gratuitous depiction of misinformation at high risk, dramatic depiction in entertainment and breaking news or opinion coverage of misinformation at medium risk, and educational treatment and news features that describe misinformation campaigns as such at low risk.
Platform rules give useful examples of harm. Google's publisher policies (opens in a new tab), under unreliable and harmful claims, bar ads on content that makes demonstrably false claims that could significantly undermine participation or trust in an electoral process, that promotes harmful health claims or contradicts authoritative scientific consensus on a current, major health crisis, and that contradicts authoritative scientific consensus on climate change. A brand can adopt those areas, narrow them or add its own.
Five steps to a misinformation clause that holds up
- Name the harm areas. List the subjects where false claims cause real damage for this brand's customers, such as health, elections or emergencies.
- Require a checkable falsehood. The clause covers claims that can be shown false with evidence, not views the brand disagrees with.
- Judge repeated behavior. Exclude sites that repeatedly publish false claims, not sites that once ran an error and corrected it.
- Name the evidence source. Say whose ratings or fact checks the team relies on, and how a site can challenge a rating.
- Keep coverage of misinformation in reach. Reporting that debunks or explains a false claim sits at low risk and stays eligible.
How do you avoid blocking legitimate debate?
The fastest way to ruin a misinformation clause is to let it drift into a list of opinions someone dislikes. Policy arguments, minority scientific views stated as such, satire and partisan commentary are not misinformation under the framework's definition, because they are not verifiably false claims. Excluding them turns a safety rule into a viewpoint rule, which costs reach and invites the charge of bias.
| Content | Example | Sample treatment |
|---|---|---|
| Verifiably false claim linked to harm | A page saying a storm shelter is closed when officials say it is open | Exclude |
| Contested policy argument | An op ed against a new power grid rate plan | Keep; judge under politics tolerance |
| Reporting on a false claim | A news story explaining how a fake aid message spread | Keep at low risk |
| Satire clearly labeled | A humor site's invented headline about a mayor | Keep if the brand accepts satire |
| Error later corrected | A local paper's wrong figure, fixed with a note | Keep; corrections are a good sign |
Before a site is excluded for misinformation
- The reviewer can point to specific false claims on the site, with the evidence that shows them false.
- The claims are linked to a harm area named in the policy.
- The pattern repeats across several pages or months, not one article.
- The site does not correct errors when shown evidence.
- The exclusion would stand if the site's politics were the opposite.
Our guide to political content covers how to set a tolerance for opinion and election coverage, which is a separate decision from misinformation and should stay separate in the policy.
Who decides whether a site spreads misinformation?
Several parties make the call, and a brand should know which one it relies on. Platforms decide for their own ad systems under policies such as Google's. Independent rating organizations assess news sites against published criteria; the Nature study drew on two of them. Fact checkers judge individual claims, and some follow the International Fact Checking Network's Code of Principles (opens in a new tab), run by the Poynter Institute, with five commitments: non partisanship and fairness, transparency of sources, transparency of funding and organization, transparency of methodology, and an open and honest corrections policy.
The final decision about where a brand's money goes belongs to the brand. In September 2025 the Federal Trade Commission (opens in a new tab) approved a final order in Omnicom's acquisition of IPG that removes the agency's ability to deny ad dollars to publishers based on their political or ideological viewpoint, except at the express and individualized direction of its advertiser clients. Whatever its scope for other companies, it shows why the brand, not an intermediary, should own and sign the misinformation clause.
What does a misinformation clause change on a real plan?
Fictional example
A backup generator dealer in hurricane season
A fictional dealer of home standby generators in Lafayette and Lake Charles, Louisiana, plans display ads for hurricane season beside storm preparation guides, local news and home improvement content in English. Its brand team fears appearing beside false storm rumors, and its previous agency had blocked every page mentioning power outages, utility rates or climate.
The new clause names one harm area, emergency information, and excludes sites that repeatedly publish false claims about storms, shelters or aid that officials have contradicted. Two small sites that spread invented evacuation orders during the last season are excluded, with links to the official corrections. Local television news, a utility rate debate on a regional business site and a weather blog's explainer on how fake aid messages spread all return to the plan.
The brand's marketing director signs the clause, and the agency records each exclusion with its evidence and a date to review it.
- One harm area named: emergency information.
- Exclusions limited to repeated, evidenced false claims.
- Debate and debunking pages restored.
- The clause signed by the brand, not the agency.
How does PlacementGraph help a brand keep misinformation off a plan?
Frequently asked questions
Is satire a form of misinformation?
Not when it is labeled and recognizable as satire. The framework's definition needs verifiably false or willfully misleading content linked to harm, and clearly labeled humor does not try to mislead. Problems start when satire is copied without its label and spreads as fact, which is a reason to judge the page where the ad runs, not the original joke.
Does one false article make a website a misinformation site?
No. The Nature study, for example, used a rater's criterion of whether a site repeatedly publishes false content. Newsrooms make errors. What separates a misinformation site is a pattern of false claims and a refusal to correct them, so a policy should look at repeated behavior and at the site's corrections record.
Should a brand avoid sites whose opinions it disagrees with?
That is a viewpoint decision, not a misinformation decision, and it should be made openly by the brand if at all. Folding it into a misinformation clause weakens the clause and invites challenge. The FTC's 2025 order on Omnicom shows regulators paying attention to viewpoint based exclusions made by intermediaries.
Are corrections a sign of a trustworthy publisher?
It can be. An open corrections policy is one of the five commitments in the IFCN Code of Principles, and a visible correction shows that the publisher checks and fixes its work. A site with no corrections on record is not shown to be more accurate; it is less transparent.
Sources
- Companies inadvertently fund online misinformation despite consumer backlash (opens in a new tab)Ahmad, Sen, Eesley and Brynjolfsson, Nature 630, June 2024
- Brand Safety Floor and Suitability Framework (opens in a new tab)Brand Safety Institute
- Google Publisher Policies (opens in a new tab)Google Publisher Policies Help
- The commitments of the IFCN Code of Principles (opens in a new tab)International Fact-Checking Network, Poynter Institute
- FTC Alters Final Consent Order in Response to Public Comments, Preventing Coordination in Global Advertising Merger (September 26, 2025) (opens in a new tab)Federal Trade Commission
- WFA discontinues GARM (opens in a new tab)World Federation of Advertisers, August 9, 2024
Researched and written by PlacementGraph Research. Read how we research, source and update these pages. Corrections: diana@placementgraph.com.