Last year, Google blocked or removed 5.1 billion ads, and restricted 9.1 billion ads with sensitive content (or limited distribution). The search engine, which is also the world's largest advertising network, suspended 39.2 million advertiser accounts, the majority of them before an ad was even displayed. Finally, 220,000 sites were targeted by a suspension measure.
When AI identifies AI threats
An impressive overall assessment, drawn up in the Ads Safety report for 2024. Google notes the rise in ads for AI-related scams. Among the most worrying threats: scams based on deepfakes of celebrities generated by generative AI tools. Stars like Tom Hanks regularly denounce the misappropriation of their image for these real-fake ads.
These stars promote, against their will, financial scams based on cryptocurrencies and other scams. In response, Google suspended more than 700,000 accounts for impersonation, leading to a 90% drop in reports of this type of fraud.
Despite the rise in malicious uses of AI, Google says these same technologies are also used for scam detection, sometimes even before user reports. The company made dozens of improvements to its machine learning models last year to strengthen detection and moderation at large scale.
But AI alone can't do everything: a group of 100 experts has been mobilized to put in place countermeasures, including an update to Google's policy on disinformation.
Source: Google (PDF)
0 Comments