Do AI Assistants Read Your Newsletter or Substack? Only the Copy That Lives on the Web

You write a newsletter every week. It goes to a few hundred people who open it, reply to it, and hire you from it. So it is fair to wonder whether ChatGPT and the other assistants read it too.

Do AI assistants read newsletters? Not the part your subscribers get. They read the copy of an issue that lives on the open web, if there is one, and only as much of it as a stranger could read.

Here is the short answer, then the five places one issue can live, the two settings that change what gets through, and how to make the writing you already do count.

Do AI assistants read your newsletter?

Only the web copy. An assistant cannot open anyone's inbox, so the email version of your newsletter is invisible to it. What it can read is an issue published as a public web page, readable without signing in, on a site that does not tell crawlers to stay away.

Most newsletter platforms publish both at once: the email goes out, and the same issue appears in a web archive. That archive is the only part of your newsletter that can reach an AI answer, which makes its settings more important than your open rate.

WHO CAN READ ONE ISSUE, AND FOR HOW LONGFIVE PLACES IT CAN LIVE

Subscribers AI assistants

Send dayFirst weekMonths laterYears later

1. The email in a subscriber's inboxSubscribers, around send day. Assistants: never.

2. A post in a private publicationSubscribers only, for as long as it is up.

3. A paid post with a free previewAssistants: the preview only, for as long as it is up.

4. A free post with AI training blockedAssistants: depends on the assistant.

5. A free public postAssistants: the whole issue, for as long as it is up.

The email is read around the day it arrives. The web copy keeps answering for as long as it stays up.

AnswerHalo answerhalo.com

CLICK TO ENLARGE

The difference in length is the point. An email is mostly read around the day it arrives, and then it sinks down the inbox. The web copy of the same issue can be found and quoted months or years later, every time a buyer asks a question it answers, so a weekly newsletter with a public archive adds fifty-two readable pages a year.

The five places one issue can live

The same issue can be invisible, partly visible, or fully visible to an assistant depending on where it lives. The table below walks through the five cases most coaches and consultants run into, what an assistant can read in each, and what to do about it.

The paid row is the one most writers misjudge. A crawler is a stranger without a subscription, so as a rule it sees what a stranger sees. Google publishes separate guidance for paywalled content precisely because a paywall changes what a crawler can read, and you should not assume your platform shows crawlers more than it shows the public.

The two settings that quietly change it

Two settings decide most of what an assistant can read of a newsletter: whether the publication is private, and whether you have told AI tools not to train on it. Both are one toggle, and both are easy to switch on without seeing the cost.

The first is privacy. Substack's help page on private publications says readers will not be able to see any posts once a publication is private, and that the welcome page may still appear in web search while no posts will. A private newsletter is a closed room. That can be exactly right for a paid community, and it means none of those posts can be quoted.

The second is the AI training toggle. Substack's page on blocking AI training says the setting tells tools like ChatGPT, Claude and Gemini not to train on your published content, that it only applies to tools that respect it, and that blocking training may limit your discoverability in tools and search engines that return AI-generated results.

Training and search are not always the same door. OpenAI's list of its crawlers separates GPTBot, which gathers content that may be used to train its models, from OAI-SearchBot, which is used to surface websites in ChatGPT's search features. How any one assistant treats your toggle is not something you can see from the settings page, which is why the table marks that row as uncertain rather than guessing.

Whose name the issue helps

An issue on a platform address helps your business only if the page itself says whose business it is. On a platform subdomain the page sits on the platform's site, so the connection to you has to be written into the issue: your name, what you do, who you do it for, and where.

Substack's own advice points the same way. It recommends using your name in your publication's address, because readers are more likely to search for your name than for a publication title they do not know, and it recommends linking the publication from your personal site. Both help an assistant join the newsletter to the person.

The same logic applies to anything you say somewhere other than your own site. It is the reason a podcast appearance only counts when it leaves readable text behind, covered in how to make your podcast and webinar appearances count.

How to make a newsletter count

Treat each public issue as a page a buyer might land on from an AI answer, not as an email that happens to be archived. That means five small habits, none of which changes how the email reads to your subscribers.

Title the issue as the question it answers. Put a two or three sentence answer at the top, before any paywall. Sign it with your name, what you do, and your city. Keep the archive public and free for the issues that answer buyer questions. And link the archive from your website, so the newsletter and the business point at each other.

If other people's newsletters already cover your field, being mentioned in the ones AI already cites can matter more than your own list. How to find those and what to send is in how to get mentioned on the sites AI already trusts.

The damaging admission: a public archive makes a newsletter readable, not recommended. Plenty of excellent newsletters are fully visible and never named, because a buyer's question is decided by everything the web says about the whole field. What the steps above remove is the reason an assistant could never name you from your newsletter at all.

To see whether you are named on your buyers' questions today, start with the free check. It asks ChatGPT 25 real buyer questions from your niche at $0 and emails a plain-English readout: your score, who got named in your place, and the one fix to start with. It runs as soon as you ask for it, and the report lands within the hour between 7am and 5pm Central, or by 10am the next morning outside those hours.