An internal Google document shows that Alphabet Inc. considered asking web publishers to allow or opt out of having their content used in AI features. However, Google decided not to offer either choice.
The memo, revealed during the company’s antitrust trial in Washington, gave publishers a way to agree or decline, which would have made the process too complex. Therefore, the plan was to quietly update Google Search without any public notice.
Bloomberg reported that the draft, written by Chetna Bindra, a product management executive at Google Search, had set a “hard red line.”
It stated that any publisher wanting its content to appear in regular search results would also have to allow Google to feed that content into AI-powered features. As the memo put it, “Do what we say, say what we do, but carefully.”
Google’s firm hold on the search market, which a federal court marked as an illegal monopoly last year, has helped it dominate the new field of AI-driven search tools.
Under Google’s own rules, material that shows up in standard results can also be tapped to train other AI-backed search products. Publishers can only block their data from those AI tools if they remove their site entirely from Google Search.
For many sites, Google’s traffic is too important to lose. The company still controls over 90% of the search market, making it the main portal to the wider web.
Many publishers have allowed Google to use their pages in AI Overviews, which supply direct answers in the search results. But by giving users what they need without clicking on any links, AI Overviews can cut into the ad revenue and sales that sites rely on.
Paul Bannister, chief strategy officer at Raptive, which represents online creators, said “It clearly shows they knew there were other options but chose the most protective one, the option that didn’t give publishers any controls at all”.
The trial in Washington completed witness testimony on May 9, with closing arguments set for later this month. Judge Amit Mehta is reviewing the recommendations by antitrust enforcers, with a decision expected in August.
Google considered several AI opt-outs but picked none
In internal slides, Google listed several middle-ground approaches. One of them was “SGE-only opt-outs,” which would let publishers block their pages from certain AI features in the Search Generative Experience without disappearing from search.
Another would let sites choose not to appear in AI Overviews, while still allowing their content to be used for training. Google also discussed letting sites block their content from grounding, the process of anchoring AI models in real sources.
In the end, Google dropped all those new choices. The slides recommended “no new controls BUT reposition publicly,” pointing publishers to the existing “no snippet” option. That setting keeps a link in search results but hides the preview text and any AI summary, making clicks much less likely.
A Google spokesperson told reporters that publishers have always had control over how their content is used in Search and AI. “This document is an early-stage list of options in an evolving space and doesn’t reflect feasibility or actual decisions,” the spokesperson said, adding that Google regularly updates its public documentation.
Bindra’s draft, written in April 2024, included notes on how to discuss the ideas and what language to avoid. It ended by saying that if the plan moved forward, the team would “work on actual language and get this out.”
One month later, in May 2024 at Google’s annual I/O conference in Mountain View, California, the company revealed what it called a “fully revamped” search experience infused with AI.
Brooke Hartley Moy, chief executive of Infactory, an AI startup that works with publishers, warned that if Google’s models get good enough, they could replace much of the work human writers do. “If Google’s models get to a point where the human element of content is diminished, then they’ve kind of signed their own death warrant,” she said.
