Google’s search dominance has been under threat ever since OpenAI unveiled its ChatGPT chatbot in late 2022. The fears of AI cutting into Google’s search business became even more real as OpenAI introduced search capabilities in ChatGPT last year, even as rivals like Microsoft’s Bing and Perplexity already provided similar solutions.
ChatGPT’s search capabilities are supposedly so great compared to Google that its CEO, Sam Altman, in a recent interaction with The Verge, said that he doesn’t use Google Search anymore.
“I legitimately cannot tell you the last time I did a Google search,” Altman said.
However, a new report by The Information has now stated that OpenAI has been using data scraped from Google Search in order to help ChatGPT answer real-time questions about issues like news, sports, and financial markets, areas where its own AI tools aren’t up to the mark yet.
OpenAI is reportedly scraping data via SerpApi—a paid web-scraping service that allows developers to extract real-time search engine results.
Apart from OpenAI, SerpApi’s client list also includes tech giants like Meta, Apple, and Perplexity. The web scraper had earlier publicly listed OpenAI on its website, but that reference has reportedly been removed now.
Prior to the report by The Information, a former Google engineer, Abhishek Iyer, had also shown how ChatGPT uses Google Search to answer real-time queries. Iyer created a few dummy web pages which appeared only in Google’s index, and later when he prompted ChatGPT about information from those pages, the chatbot was able to access that information.
Google had rejected OpenAI’s request for its search index:
Officially, OpenAI has previously said that its search functionality on ChatGPT is powered by its own web crawler, along with data from Bing and licensing partnerships with news and content publishers. However, during the Google anti-trust trial last year, ChatGPT head Nick Turley had admitted that the company had reached out to Google for its search index but that request was denied.
Turley had also confirmed that it had “significant quality issues” with Microsoft’s search results and “it was at best a near-term solution.”
“Having access to the data that underlies Google’s index, the content or signals, would accelerate the development of our own index,” Turley had said last year.
