Saturday, July 13, 2024

Google Releases 2023 Ads Safety Report, AI Becomes Key

2023 was a turning point for the digital advertising industry. With

the rise of generative AI, the online advertising ecosystem underwent significant changes and faced new security challenges. As the world’s largest advertising platform, Google detailed in this year’s “Google Ads Safety Report” how they leveraged the latest AI technology to strengthen their fight against malicious ads and fraudulent content.The report shows that Google blocked or removed a total of 5.5 billion bad ads in 2023. Among them, 1.04 billion ads were blocked for abusing the ad network, 206.5 million were removed for misrepresentation, and 273.4 million violated financial services-related policies.The report points out that Large Language Models (LLMs) were the key technology Google used to enhance ad security last year. Thanks to the powerful semantic understanding and reasoning capabilities of LLMs, Google was able to identify and block non-compliant ads with greater precision and on a larger scale. Compared to traditional machine learning models, LLMs showed advantages when facing constantly changing financial fraud tactics and could promptly identify emerging fraud patterns. Google stated that over 90% of publisher page-level enforcement last year was initially screened through machine learning models, including LLMs.In 2023, Google took action against 2.1 billion web pages, with 1.8 billion pages being penalized for sexual content and another 104 million pages being flagged for dangerous or derogatory content.The report also mentions that as scammers employ more sophisticated techniques such as deepfakes, Google has set up dedicated rapid response teams to optimize algorithms and remove fake celebrity scam ads on a large scale. In November last year, Google also launched the “Limited Ads Serving” policy, imposing more restrictions on new advertising clients without a good reputation record to reduce the risk of fraud.It is worth mentioning that in the major elections last year, Google implemented stricter transparency requirements for political ads and took the lead in introducing new regulations for election ads containing synthetic content. Google verified more than 5,000 new election ad publishers last year and removed 7.3 million election ads that did not complete identity verification. This helps voters obtain sufficient information to make informed decisions.Overall, this report reflects the increasingly important role AI technology plays in the field of ad security. Looking ahead, Google says it will continue to invest in policy development, violation detection, and enforcement actions to address potential challenges. In 2023, Google made 31 updates to its advertising and publisher policies and launched the “Ads Transparency Center” to make it easier for users to search and understand all ads from verified advertisers. Google emphasizes that collaborating with the industry and giving users and partners more control is key to maintaining a healthy advertising ecosystem. As a tech giant, Google will inevitably take on more responsibility in the AI-driven advertising ecosystem, and its future is worth watching.