EFF: Google Summaries Deepen Lock-In

The Electronic Frontier Foundation's Deeplinks blog on November 4, 2026, ran a piece by Katharine Trendacosta with a blunt title: “Google Summaries Aren't About AI, They're About Lock-In.” The article's central argument is that the search giant's AI-generated summaries are less a technical milestone than a mechanism for keeping users inside Google's ecosystem. The post's byline, headline, and publication date are the entirety of what the source text makes explicit about the author and timing; no further biographical or organizational detail about the author is available from the source.

A headline that names the real target

The source's headline is itself the thesis. Trendacosta's post is framed around the claim that the AI label attached to Google's summaries obscures the product's more consequential effect: tying users more tightly to Google's search results and away from the open web. The article does not present new empirical data or a formal study; it is a short Deeplinks commentary whose argument rests on the headline's framing.

The source includes no subheadings, quoted sources, or statistical tables. What it does include is a clear editorial position, attributed to Trendacosta, about what Google's AI summaries are for. The EFF's Deeplinks blog is listed on the site under its Issues and About navigation, but the post itself does not elaborate beyond its title and byline in the text supplied.

The problem with summarization

The concern the headline points at is straightforward: a search engine that answers a user's question directly, inside the results page, reduces the user's need to click through to the original source. If the summary is the destination, the sites that produced the underlying information become less visible and less visited. The source post makes this point through its title's contrast between “AI” and “Lock-In” — suggesting the summarization feature is better understood as a retention mechanism than as a neutral new capability.

No data is presented in the post to quantify that effect. There are no click-through figures, no traffic charts, and no case studies in the source. The argument is structural: the feature changes where the answer lives, and where the answer lives shapes where the user stays.

Why “lock-in” is the operative phrase

The EFF post's chosen frame is lock-in — the condition in which a user's accumulated dependence on a platform makes leaving costly. In the search context, that dependence is built from habit, default settings, and the convenience of getting an answer without leaving the page. The post's headline asserts that Google's AI summaries should be read through that lens rather than through the lens of technological novelty.

The source text does not elaborate on the mechanics of lock-in or offer a definition beyond the headline's use of the term. It does not claim that Google introduced the feature with lock-in as a stated goal, nor does it present internal documents or statements from the company. The post is an opinion piece whose argument is carried by its title and by the byline of a named EFF staffer.

What the source does not say

The supplied text of the article contains no direct quotations, no statistics, and no named Google spokesperson. It does not describe a specific incident, a regulatory filing, a lawsuit, or a product change date. There is no mention of competitors, market share, referral traffic trends, historical accusations against Google, or policy prescriptions. Any coverage that adds those elements would be drawing on material not present in the source, and this article does not do so.

The piece also does not present a timeline of Google's AI summary rollout or describe how the feature is technically implemented. It does not state which countries or user segments have access to summaries, and it does not quantify how often summaries appear in search results. The source is a brief opinion post, and its factual content is limited to the existence of Google's AI summaries, the EFF's characterization of them as a lock-in mechanism, and the identity of the author.

A named author, a short argument

The post is attributed to Katharine Trendacosta. The source text does not include a job title for Trendacosta beyond the byline format used on the Deeplinks blog, and this article does not invent one. The EFF's site lists its staff under a People page, but the source text itself does not provide additional biographical context.

What the source does establish is that the EFF's Creativity & Innovation issue area is tagged on the post. That tag aligns with the subject matter: a debate about how search results are presented and who captures the value of the underlying information. The post does not go into detail about the Creativity & Innovation program or its other work.

The editorial position, stated plainly

The source's position is that Google's AI summaries are not primarily an AI story. The title calls them a lock-in story. That is the whole of the argument as presented in the supplied text — no supporting quotes, no data, no elaborations. Readers who want the full post can find it on the EFF's Deeplinks blog at the URL cited in the source metadata.

Because the post is short and argumentative rather than investigative, it does not claim to have discovered anything new about Google's product. It offers a reinterpretation of a well-known feature, and it asks readers to see the feature through a competition-and-dependence lens rather than a technology lens.

What this means for readers

The practical implication of the EFF's argument, if one accepts it, is that the convenience of an in-page answer is not free. Every question answered inside a search results page is a question that did not send a reader to the site that produced the answer. For publishers, that could mean fewer visits from search; for users, it could mean more of their information diet is mediated by a single company's summarization choices. The source post does not quantify those effects, and this article does not either.

For businesses and site operators, the EFF's framing suggests that the visibility they once derived from ranking in search results may depend increasingly on how a summary layer chooses to represent them. The post does not offer guidance on how to adapt, and it does not name any specific mitigation. Its value to readers is as a prompt: to notice when a feature is presented as a neutral technical improvement but also changes where attention and traffic flow.

For policymakers and competition regulators, the post's title raises a question without answering it: whether lock-in effects from AI summaries fall within existing frameworks. The source does not make a policy recommendation or cite any rule. It simply reframes the feature. Whether that reframing gains traction is not something the source text can tell us.