Google uses AI detection to pause more than 39 million ads suspected of being scammed

On Wednesday, Google showed its latest data on the results of its fight against fraudulent ads, and in 2024 Google suspended about 39.2 million advertiser accounts on the platform, more than tripling the number of blocks last year. Google uses a large language model (LLM) to detect details such as impersonating business names and illegal payment methods, and can suspend most fraudulent accounts immediately before the ads go out. Last year, Google introduced more than 50 LLM enhancements and a team of human experts to dramatically improve security across Google's platforms.

Google uses AI in conjunction with a panel of experts to combat fraudulent ads

Alex Rodriguez, general manager of advertising safety at Google, said at a conference that while AI large language models are important, humans are still needed to be involved in the process. The executive told reporters that a team of more than 100 experts has been established within Google, including members of the ad safety team, the trust and safety department, and researchers at DeepMind. They analyze Deep Fake scam ads that impersonate public figures and develop countermeasures.

Google suspended more than 39.2 million suspicious ad accounts

Last year, Google rolled out technical countermeasures and updated more than 30 advertising and publisher policies, which helped suspend more than 700,000 ad accounts with suspected violations and a 90% reduction in ads involving Deep Fake scams. In the U.S. alone, 39.2 million advertiser accounts were suspended and 1.8 billion offending ads were taken down, involving fraudulent advertiser practices including trademark abuse, healthcare, personal advertising, and misrepresentation.

Online ad scam accounts are suspended in violation of Google policies

India is the world's most populous country and the second largest online market after China. Google said 2.9 million accounts were suspended in India last year, and Google removed 247.4 million ads in India, with the top five violations related to financial services, trademark abuse, ad network abuse, personalized advertising, and gambling and gaming.

Google says 5 million of all suspended advertiser accounts have been suspended for fraud-related violations.

Google told TechCrunch that it blocked 5.1 billion ads and removed 1.3 billion pages for all of last year. This compares to more than 5.5 billion ads blocked and 2.1 billion publisher pages acted upon in 2023. The drop in numbers shows an improvement in fraud prevention. Fraud damage can be reduced by improving early detection of malicious accounts and suspending campaigns.

Mass suspension of campaigns sometimes raises concerns about whether Google is fairly enforcing the rules, and Google also provides a human-reviewed appeal process to ensure appropriate action is taken.

This article by Google using artificial intelligence to detect the suspension of more than 39 million ads suspected of scams first appeared in Chain News ABMedia.