Conspiracy theories and misinformation about vaccines dominate Instagram's search results, and are made more visible by the app's recommendation algorithms
Instagram's efforts to curb health misinformation have done little to stem the flow of conspiracy theories and misinformation about vaccines. Tweets: @sachabaroncohen , @alfiekohn , @chelseaclinton , and @sykescharlie . Thanks: @jpmanga Tweets: Sacha Baron Cohen / @sachabaroncohen : This is terrifying. What if millions of people refuse a vaccine for coronavirus because of lies they've seen on social media? It's time to hold Facebook and YouTube liable for the content they spread and the deaths they facilitate! https://www.nytimes.com/... Alfie Kohn / @alfiekohn : “Millions of lives...may depend not just on producing a vaccine, but on persuading people to accept it” in the face of “a well-oiled anti-vaccine movement that has already polluted the air with misinformation & conspiracy theories”: https://www.nytimes.com/... Chelsea Clinton / @chelseaclinton : Yes. Sign me up: “We need to mobilize a pro-vaccine movement that is as devoted, as internet-savvy and as compelling as the anti-vaccine movement is for its adherents.” #VaccinesWork https://www.nytimes.com/... Charlie Sykes / @sykescharlie : “What if we get a Covid-19 vaccine and half the country refuses to take it?” https://www.nytimes.com/... Thanks: @jpmanga
Context & Ripple Effects
This is the third documented failure of the same pledge. In March 2019 Facebook announced it would curb vaccine misinformation by demoting anti-vaccine Pages and groups, halting their recommendations, and rejecting their ads (the demotion-and-ads crackdown). By May 2019 reporters were still finding anti-vaccination posts and #vaccineskill tags prominent in Instagram search (two months into that policy), and in February 2020 HuffPost found search unchanged a year after Facebook said such content wouldn't appear on Explore or hashtag pages.
The Engadget finding closes the loop: the problem isn't only residual content but active amplification — Instagram's recommendation algorithms make conspiracy theories more visible in search. That matters acutely because the reporting lands mid-pandemic, with Sacha Baron Cohen and others citing the risk that people refuse a future COVID-19 vaccine because of what they saw on social media, and calling to hold Facebook and YouTube liable. Bloomberg's later reporting on Facebook's "imminent harm" standard — abandoned in February 2021 after it let dangerous COVID lies spread — shows where this pressure eventually pushed the company's policy line.
First-order effects
- Users searching vaccine topics on Instagram get conspiracy content ranked up by recommendation algorithms, meaning Facebook's own 2019 levers — Explore, hashtag pages, ad rejection — are demonstrably not touching the search-and-recommend surface where this spreads.
- Public figures like Sacha Baron Cohen, Chelsea Clinton, Alfie Kohn, and Charlie Sykes seize on the finding to press a liability argument against Facebook and YouTube specifically over vaccine refusal deaths.
Second-order effects
- Facebook's stated policy framework comes under direct strain: a moderation regime built on ranking tweaks rather than removal keeps producing these audits, feeding the case for legal liability that Cohen's statement crystallizes — pressure that precedes the company's later reversal on its "cause imminent harm" threshold.
- Each recurring audit (CNN 2019, HuffPost 2020, Engadget 2020) hands regulators and advertisers a documented pattern rather than a one-off incident, raising the cost for Facebook of framing the issue as solved after each announcement.
Third-order effects
- The pattern points toward a structural conclusion: ranking-based commitments don't neutralize recommendation-driven amplification, so durable fixes shift from platform self-regulation to external accountability — liability frameworks and regulator-set standards for what surfaces in search and recommendations.
- If health-misinformation visibility is ultimately governed by outside rules rather than internal policy cycles, platform moderation becomes a compliance function measured by auditable search outcomes instead of by policy announcements.
The trend: Platform self-regulation of health misinformation keeps losing to recommendation-driven amplification, steadily shifting the debate toward legal liability and external regulation of what algorithms surface.