Google allowed advertisers to target racist and bigoted keywords, suggested more similar keywords, and allowed campaigns using those keywords to run
Google, the world's biggest advertising platform, allows advertisers to specifically target ads to people typing racist and bigoted terms …
Context & Ripple Effects
This BuzzFeed report caught Google's ad machine doing the opposite of what its brand-safety marketing promised: advertisers could buy racist and bigoted keywords outright, and Google's own keyword tool suggested more of them. It became the reference point for a decade of follow-on reporting on the same enforcement gap.
That arc runs through Google's Ad Exchange demonetizing pages that merely mention race and LGBTQ rights in 2020, the political-ads targeting loophole Bloomberg exposed in 2019, the secret YouTube hate-term blocklist that failed more than two-thirds of the time, and Wired's 2025 finding that Display & Video 360 still lets users target sensitive personal data despite rules against it.
First-order effects
- Advertisers buying these keywords get their campaigns served against search queries Google's own tools actively recommend, meaning the platform is both marketplace and accelerant for bigoted targeting.
Second-order effects
- Brand-safety pressure pushes Google toward blunt automated fixes — the same machinery later shown to demonetize legitimate race and LGBTQ coverage, shifting revenue pain onto publishers rather than bad actors.
Third-order effects
- A recurring pattern of stated policy versus actual enforcement across keywords, politics, and audience data gives regulators and litigators a documented track record to argue that self-policing at ad-platform scale does not hold.
The trend: Google's ad business keeps oscillating between under-enforcement against harmful targeting and over-enforcement against legitimate content, and each investigative data point strengthens the case for external oversight of ad-platform policy execution.