Afternoon BriefAI Search & Discovery

Google AI Mode and ChatGPT Both Moved Citations to the Bottom. Here Is What They Are Actually Telling You.

Google AI Mode and ChatGPT are both testing bottom-card citations. Two engines, same week, same message: the answer is the product and your source link is the appendix. What founders need to know about source authority vs citation count.

Jaxon Parrott
Jaxon ParrottJul 23, 2026

Two things happened this week that most marketing teams will ignore. Google AI Mode started anchoring citation cards to the bottom of its answers. ChatGPT started hiding citations behind a click. Two different engines. Same structural decision. Same message: the answer is the product. Your citation link is the appendix.

I have spent eight years building a company around the idea that what machines extract from your content matters more than where they link to it. This week, both platforms confirmed that thesis in production code.

What Google and ChatGPT Changed This Week

On July 22, Sachin Patel spotted Google AI Mode testing a new citation format: instead of inline source links woven into the answer, citations now anchor to card panels at the bottom. The same day, Brodie Clark reported that ChatGPT was testing click-activated overlay boxes where citations appear only when a user requests them.

Two unrelated engineering teams. Same architectural choice. Same week.

Google's version uses an anchor mechanism that directs users down to a citation card panel. ChatGPT's version keeps the initial answer clean and surfaces sources only on demand. Google has been iterating on this trajectory since at least March 2026, when it began testing "Ask About" overlays that let users ask follow-up questions about a source without ever visiting the original page.

The convergence is not coincidence. It is two platforms arriving at the same conclusion: users come for the answer. Sources are supporting evidence, not the main event.

The Citation Was Never the Asset

Here is the uncomfortable read most teams will skip past.

When your brand name appears as a bottom-card citation in Google AI Mode or behind a click gate in ChatGPT, the user almost certainly will not click it. Google's "Ask About" feature already lets users dig into a source without leaving the AI answer. ChatGPT's overlay keeps the user inside the conversation. In both designs, the citation exists for the engine's credibility. Not for your traffic.

But here is what those engines still need from you: the content they are restructuring their answer around.

An analysis of 8,500 AI-generated answers across four platforms found that only 17% to 54% of AI Overview citations come from the top 10 organic results, depending on query type. Moz's review of 40,000 keywords showed just 12% of Google AI Mode citations match top-10 organic results. The overlap between Google rankings and AI citations collapsed from 70% in 2024 to under 20% by May 2026.

The engines are not selecting sources the way Google Search does. They are selecting the content they can extract the most useful answer from. That is a completely different competition.

What Actually Gets You Into the Answer

A meta-analysis of 54 studies found that brand mentions across the web correlate roughly 3x more strongly with AI citation than backlinks do. Content updated within 60 days is 1.9x more likely to be cited. And 44% of citations are extracted from the first 30% of the text.

Three things determine whether your brand is in the AI answer body or filed away in a bottom card nobody scrolls to:

1. Source dependence, not citation count. The engine needs your content to construct its answer. That means original data, named frameworks, proprietary terminology, or evidence the engine cannot find anywhere else. A brand with unique research the engine relies on is structurally more valuable than a brand with ten links in generic "best of" lists.

2. Passage extractability. If 44% of citations come from the first 30% of the text, then burying your answer under 800 words of setup makes you invisible to the retrieval system. The engines are pulling the specific passage that answers the user's question. They are not evaluating your domain authority. They are evaluating your paragraph.

3. Cross-platform earned authority. Only 16% of domains cited by one AI engine are also cited by another. Each engine retrieves from different pools. ChatGPT leans on Wikipedia and Bing-indexed content. Perplexity over-indexes on YouTube and Reddit. Google AI Mode cites its own properties at 17.42%. If your authority signal only lives in one ecosystem, the other engines do not know you exist.

Why This Is a Machine Relations Problem, Not an SEO Problem

SEO taught brands to optimize for where they rank. Machine Relations exists because ranking position no longer predicts citation. The data is definitive: 88% of AI-cited URLs come from sources that do not rank in the traditional top 10.

When Google and ChatGPT both move citations to the bottom, they are confirming what the architecture has always rewarded: being the source the engine structurally depends on. Not the source that got a link. The source the engine cannot build a credible answer without.

The brands that will win this are the ones building source architecture: original research, earned third-party mentions, passage-level answers that are the best available response to a specific question. Everything else becomes a footnote.

Literally.

FAQ

Are bottom-card citations a permanent change or just a test?

Both Google and ChatGPT are actively testing these formats. Google has been iterating on AI Mode citation display since March 2026, adding "Ask About" overlays and anchor mechanisms. The direction is consistent across both platforms: source attribution is being separated from the answer body. Whether the exact UI persists, the architectural intent is clear and unlikely to reverse.

Does a bottom-card citation still help my brand?

It depends on what the citation represents. If the engine paraphrased your original data or framework in the answer body, the brand value is in the engine relying on you, not the link placement. If you are one of nine generic sources in a bottom card, the value is close to zero. The question is not where the citation appears. The question is whether the engine needed your content to construct its answer.

How do I know if an AI engine depends on my content?

Run the same query across ChatGPT, Google AI Mode, and Perplexity. If the AI answer includes specific claims, data points, or frameworks that originate from your content, the engine depends on you regardless of where your link sits. If your brand appears in a citation card but none of your original thinking is in the answer, you are a footnote. Not a source.