Overview
- Internal documents disclosed in a U.S. antitrust trial show Google used publisher data to train AI models without explicit consent, opting against granular opt-out options.
- Google relied on its outdated 'no snippet' setting as the sole control for publishers, which negatively impacts site visibility and traffic.
- The AI Overviews feature, launched in 2024, has diverted user clicks from publisher sites, causing significant declines in traffic and revenue for many content creators.
- Judge Amit Mehta is evaluating U.S. Justice Department proposals that could require Google to implement model-by-model and product-by-product opt-outs for publishers.
- Google executives, including Liz Reid, defended the decision in court, citing technical complexity and high costs as reasons for avoiding multiple opt-out layers.