Google automates ad safety

Google is moving more ad safety checks into automated systems that screen campaigns earlier, before many ads ever reach review queues. (blog.google.com) In its 2025 Ads Safety Report, published April 16, 2026, Google said it blocked or removed more than 8.3 billion ads and suspended 24.9 million advertiser accounts in 2025. The company said Gemini-powered tools stopped more than 99% of policy-violating ads before anyone saw them. (blog.google.com) Google also said it restricted 4.8 billion ads, actioned 1.3 billion publisher pages, and took action on more than 220,000 publisher sites in 2025. Scam enforcement accounted for 602 million blocked or removed ads and 4 million suspended accounts. (services.google.com) The shift changes where humans spend time. Instead of catching obvious violations one ad at a time, analysts are more likely to handle edge cases, review false positives, and work through approval delays when legitimate campaigns get flagged. (contentgrip.com) Google describes the newer system as one that looks at broad patterns — including account behavior, payment signals, business impersonation, and coordinated abuse — rather than relying mainly on keywords or fixed rules. The company said that helped it identify scam networks earlier in the setup process. (services.google.com) That approach arrives as Google is also adding more automation on the advertiser side. Google said Ads Advisor and Analytics Advisor began rolling out to English-language accounts in December 2025, offering automated recommendations and campaign guidance inside its ad tools. (blog.google.com) Google has been pushing “agentic” ad tools since 2025, including systems that can suggest keywords, generate assets, and recommend fixes under marketer supervision. In a March 25, 2026 post, the company said advertisers should use those tools to review readiness, monitor recommendations, and decide when to accept or reject changes. (blog.google.com) The company says the automation is also reducing mistakes. Google said new large language models cut false account suspensions by 80% during a pilot for suspicious payment activity, letting teams focus on more complex investigations. (services.google.com) Not everyone reads the numbers the same way. TechCrunch reported that blocked ads jumped from 5.1 billion in 2024 to 8.3 billion in 2025 even as suspended accounts fell, a pattern that raised questions about whether Google is getting better at stopping bad creatives early or simply acting less often at the account level. (techcrunch.com) For advertisers, the practical effect is less about writing around policy after a rejection and more about getting campaigns “ready to run” before launch. As Google moves safety checks upstream, the bottlenecks shift from manual review to diagnosis: why an ad stalled, which signal triggered scrutiny, and whether an account is trusted enough to clear automatically. (searchengineland.com)