In a small experiment, Facebook approved seven scheduled ads with content that violated company rules about COVID-19, indicating flaws in automated ad screening
Kaveh Waddell / Consumer Reports :
Context & Ripple Effects
This experiment lands a month after Facebook announced it was temporarily banning ads and listings for medical face masks as part of its COVID-19 response — so the seven approved ads show the enforcement layer lagging the policy announcement. It also extends a documented pattern: since at least 2018, Facebook's blend of AI and human ad reviewers has been criticized from both directions, snaring legitimate news and publishers while missing rule-breaking content.
Consumer Reports' small scheduled-ad test is the same methodology researchers would later run against TikTok, Facebook, and YouTube with voting misinformation in 2022, where Facebook again approved a significant share — making this 2020 finding an early data point in a recurring audit genre rather than a one-off.
First-order effects
- Ads that violate Facebook's stated COVID-19 rules can enter the paid distribution pipeline unreviewed, meaning the company's pandemic-era ad policies are only as strong as its automated screening at the moment of scheduling.
Second-order effects
- Advertisers and publishers caught in the opposite failure mode — legitimate content wrongly rejected by the same screening stack — have fresh evidence that the system errs in both directions, strengthening the case researchers made that [[a:951661|moderation policies alone don't fix the problem because targeting and algorithmic optimization are the core issue]].
Third-order effects
- If small researcher audits keep catching what automated screening misses across platforms, ad review shifts toward external verification and regulator scrutiny of platform claims about enforcement — the pattern the 2022 cross-platform test confirmed two years later.
The trend: Platform ad moderation is being progressively audited by outside researchers whose repeated experiments expose a gap between published rules and automated enforcement.