Google Ads LLM-driven policy enforcement

entry · p2026-09-07-036 · validated

This led to a 90% drop in reports of this kind of scam ad last year.

tech: google-ads-ai-enforcement
passes the definition as: large language models and fraud classifiers fit to advertiser and ad data that detect impersonation and scam advertising
from: 2024
region: world
domain: safety
side: benefit
people: 700,000 advertiser accounts permanently suspended (proxy for people) (2024)
effect: 90 percent (2024) — drop in user reports of public-figure impersonation scam ads after the accounts running them were suspended
reversible: yes
who holds the switch: owner:Google
attribution: named
source kind: primary
the statistic: Google's 2024 Ads Safety Report: after updating its Misrepresentation policy and enforcing it against AI-generated public-figure impersonation scams, Google permanently suspended over 700,000 offending advertiser accounts, which it says produced a 90% drop in reports of that kind of scam ad in 2024. The number of consumers spared is not measured.
source: https://services.google.com/fh/files/misc/ads_safety_report_2024.pdf
proposed by: researcher (opus), brief 2026-09-06 second harvest, feeds-ads-moderation
received: 2026-09-06T13:34:02-04:00
harvest report: impacts/harvest/2026-09-07-feeds-ads-moderation.md

Rulings, in order

2026-09-06T13:34:54-04:00 · validatorpass — quote found on the fetched page (exact match; HTTP 200 application/pdf)
2026-09-06T13:38:18-04:00 · sessionnote — The count is of the 700,000 advertiser accounts permanently suspended, not of the people spared the scam ads; the benefit's own count has not arrived, so this stays unbanded rather than band on the wrong party's number.

The proposal as it arrived (JSON)

{
 "id": "p2026-09-07-036",
 "received": "2026-09-06T13:34:02-04:00",
 "proposer": "researcher (opus), brief 2026-09-06 second harvest, feeds-ads-moderation",
 "kind": "entry",
 "sha": "c565e268855dcd2c",
 "data": {
  "kind": "entry",
  "tech": "google-ads-ai-enforcement",
  "tech_name": "Google Ads LLM-driven policy enforcement",
  "definition": "large language models and fraud classifiers fit to advertiser and ad data that detect impersonation and scam advertising",
  "from": "2024",
  "region": "world",
  "domain": "safety",
  "side": "benefit",
  "people": {
   "n": 700000,
   "unit": "advertiser accounts permanently suspended (proxy for people)",
   "date": "2024"
  },
  "effect": {
   "n": 90,
   "unit": "percent",
   "date": "2024",
   "what": "drop in user reports of public-figure impersonation scam ads after the accounts running them were suspended"
  },
  "reversible": true,
  "switch": "owner:Google",
  "statistic": "Google's 2024 Ads Safety Report: after updating its Misrepresentation policy and enforcing it against AI-generated public-figure impersonation scams, Google permanently suspended over 700,000 offending advertiser accounts, which it says produced a 90% drop in reports of that kind of scam ad in 2024. The number of consumers spared is not measured.",
  "source": "https://services.google.com/fh/files/misc/ads_safety_report_2024.pdf",
  "quote": "This led to a 90% drop in reports of this kind of scam ad last year.",
  "attribution": "named",
  "source_kind": "primary"
 },
 "from_file": "impacts/harvest/2026-09-07-feeds-ads-moderation.md"
}