Google Utilizes AI to Combat Harmful Advertisements, Halting 39.2 Million Fraudulent Accounts

Google is intensifying its initiatives to tackle malicious advertising through the use of artificial intelligence, yielding impressive outcomes. According to its newly released 2024 Ads Safety Report, the tech powerhouse suspended 39.2 million fraudulent advertiser accounts in the previous year as part of its continuous goal to safeguard users from harmful content.

The report showcases Google’s application of advanced Large Language Models (LLMs) to identify and thwart ad fraud. Since 2023, the company has made 50 updates to its LLMs, enabling them to recognize emerging threats, detect abuse patterns, and distinguish between legitimate businesses and scams with significantly less data than prior models necessitated. These AI technologies are adept at identifying warning signs such as business impersonation and illicit payment methods.

Due to these advancements, Google blocked or eliminated an astonishing 5.1 billion ads in 2023. A considerable number of these ads were flagged for attempting to “exploit the ad network” — a phrase that encompasses strategies like circumventing Google’s review processes, employing bait-and-switch methods, or distributing malware. Additional infractions included trademark violations and the inappropriate use of personalized advertising, especially ads targeting users based on sensitive attributes like personal challenges, identity, beliefs, or sexual orientation — all of which contravene Google’s advertising regulations.

Alongside combating traditional ad fraud, Google is also addressing the emergence of AI-driven scams. The rise of deepfakes — hyper-realistic fabricated videos and audio — has facilitated impersonation of public figures by malicious actors. In 2023, actor Tom Hanks’ likeness was utilized in a misleading medical advertisement, while Scarlett Johansson pursued legal action against an app that misappropriated her image and voice without authorization. In reaction, Google intensified its scrutiny of advertisers utilizing AI-generated material to falsely imply celebrity endorsements, suspending over 700,000 accounts engaged in such practices. This initiative resulted in a 90% reduction in user reports concerning these scams.

Overall, Google blocked or removed 415 million scam-related advertisements last year. The company’s ad safety team highlighted that the majority of the fraudulent accounts were deactivated before users ever encountered their ads — a proactive strategy that emphasizes the magnitude of the threat and the role of AI in maintaining safe digital environments.

Given the vast amount of harmful content circulating online, it’s concerning to reflect on the types of ads that never reached users’ screens — largely owing to Google’s advancing AI safeguards.