Sources: OpenAI has been partially using Google search results scraped by a startup called SerpApi for ChatGPT responses on current events like news and sports
As it tries to unseat Google, OpenAI is relying on search data from an unlikely source: Google.
Context & Ripple Effects
OpenAI’s search effort has moved from a reported plan to build a cited web-search feature to the limited SearchGPT rollout and later broader ChatGPT search access. This report adds an important detail about the upstream data path behind timely answers.
The competitive premise has long been that conversational answers could pressure Google’s search business, a threat that reportedly prompted a Google “code red” response. Reliance on results obtained through an intermediary complicates the picture of OpenAI as an independent search challenger.
First-order effects
- OpenAI’s current-events answers are, at least in part, dependent on SerpApi-delivered Google result data rather than solely on a first-party search index or direct publisher retrieval.
- SerpApi becomes a consequential intermediary in ChatGPT’s news and sports-answer pipeline, while Google’s results retain value inside a rival’s product even when users do not visit Google directly.
Second-order effects
- OpenAI faces greater operational and commercial exposure to the availability, terms, and quality of an intermediary’s Google-derived data; building the planned cited search experience does not by itself remove that upstream dependency.
- The report sharpens incentives for answer-engine rivals to secure durable direct crawling, indexing, and publisher-data arrangements, rather than treating search-result feeds as a stable substitute.
Third-order effects
- If answer products continue to rely on a small number of web-index and results providers, competition may shift from interface innovation toward control of retrieval infrastructure and the contracts that govern access to it.
- This also strengthens the case for clearer publisher and platform rules around how search-derived material can be reused in generated answers, though the report alone does not establish how Google or publishers will respond.
The trend: AI answer engines are competing for users while revealing that independent, current-information retrieval remains a strategic bottleneck.