AI News
28 Jul 2026
Read 9 min
Detect AI in Substack newsletters with 5 simple checks
Detect AI in Substack newsletters fast, so you can spot AI-written posts and trust real authorship.
Want quick ways to detect AI in Substack newsletters? Substack now includes a Pangram scanner that estimates how much of a post, note, or comment is AI-written. Pair the built-in scan with simple reading checks: look for an AI author’s note, verify sources, and spot patterns that signal machine-made text.
Substack just made it easier to judge what you read. The platform has integrated Pangram, an AI-writing detector, so readers can scan content and see a human vs AI estimate. This move may feel risky in the short term, but it aims to rebuild trust by making creation methods more clear. Writers can also add an optional AI author’s note that explains how they used tools. Substack says the goal is not to punish AI use, but to encourage honest “how I make this” disclosures and stronger editing habits.
Open a post, note, reply, or comment (over 100 characters).
Use the Pangram scan option to see a human vs AI estimate.
Treat the score as a signal, not a verdict. Follow up with the checks below.
If you are a publisher, you can run Pangram on drafts before you hit publish.
Many writers will add an AI author’s note to explain their process.
Clear notes tell you if AI helped with outlining, editing, or drafting.
Transparent process statements build trust even when AI is used.
Human writing tends to include specific places, dates, numbers, and names.
Watch for personal moments: what the writer saw, did, or felt.
AI often leans on vague claims, generic praise, and broad life advice.
Sudden tone shifts between sections can hint at pasted AI passages.
Click links. Do they support the point? Do they exist?
Check quotes. Are they accurate and placed in context?
Note how data is used. Real analysis explains how numbers were found.
AI-generated text may include confident but flimsy or mismatched citations.
Look for looping transitions like “In conclusion,” “Moreover,” “Furthermore” every paragraph.
Watch for tidy, five-sentence blocks that repeat the same rhythm.
Flag lists that say a lot but add no new facts or angles.
Scan notes and comments too; Pangram works there and can reveal formulaic replies.
These five steps help you detect AI in Substack newsletters even when you do not rely only on a detector score.
Add the optional AI author’s note. List which tools you used and where they helped.
Share what was 100% human: reporting, interviews, and editing decisions.
Rewrite AI text to match your tone. Add personal detail and original examples.
Replace generic transitions with concrete facts or new angles.
Verify every claim, number, and citation. Do not trust machine-made references.
Link to primary sources when possible.
Run Pangram on your draft. If it flags too much AI, revise more deeply.
If Pangram mislabels your work after publishing, you can report and remove incorrect scans.
It estimates, not proves. AI detectors can make mistakes, especially on short text.
It works above 100 characters, so very short posts may not scan.
It is best used with human judgment. Read closely and use the five checks.
It invites honesty. Clear disclosure often matters more than a raw score.
Open the Substack app and scan a post with Pangram.
Read any AI author’s note to learn how tools were used.
Apply the voice, source, and structure checks.
Skim comments and replies. Scan them too if needed.
Decide: subscribe, skim, or skip based on trust and clarity.
The new tools make it easier to detect AI in Substack newsletters without guesswork. When readers combine the scan with simple reading habits, they can enjoy more human ideas and better writing, no matter how much software was involved.
How to detect AI in Substack newsletters: 5 simple checks
1) Use the Pangram scan inside the Substack app
2) Look for the AI author’s note or “How I make this”
3) Listen for a steady voice and lived detail
4) Verify sources, links, quotes, and examples
5) Scan structure for repetition and stock phrasing
Why Substack is adding detection now
Substack joins a wave of platforms that label AI-made content. Social apps already tag AI images and videos. Music services label and sometimes reduce reach for AI-heavy tracks. By adding Pangram, Substack aims to keep “AI slop” out of feeds and let readers decide what mix of human and AI they accept. Substack’s CEO has said software should handle the easy parts and people should handle the hard part: the idea that matters. The new tools push in that direction. Writers get help disclosing methods. Readers get a simple way to check. Over time, that can raise the floor on quality.Tips for writers who use AI responsibly
Disclose clearly
Edit for voice
Check facts and links
Pre-scan and correct
What the detector can and cannot do
Reader workflow you can use today
For more news: Click Here
FAQ
Q: What is Substack’s new Pangram scanner and how does it help detect AI in Substack newsletters?
A: Substack integrated the Pangram AI-writing detector into its app so readers can scan posts, notes, replies, and comments to get an estimate of how much content was written by a human versus AI. The feature shows a human vs AI estimate for items over 100 characters and is intended as a signal, not a definitive verdict.
Q: How do I run a Pangram scan on a Substack post or comment?
A: Open the post, note, reply, or comment in the Substack app (over 100 characters) and choose the Pangram scan option to see a human vs AI estimate. Publishers can also run Pangram on drafts before publishing and report scans they believe are mistakes.
Q: What does a Pangram score mean and how should I interpret it?
A: Pangram provides an estimate indicating the likely mix of human and AI authorship, but it does not prove origin and can make mistakes. Treat the score as a signal and follow up with manual checks like looking for an AI author’s note and verifying sources before drawing conclusions.
Q: What manual checks can I use to detect AI in Substack newsletters besides the scanner?
A: Use the Pangram scan alongside manual checks: look for an AI author’s note or a “how I make this” disclosure, listen for steady voice and lived detail, verify sources and quotes, and scan structure for repetition or stock phrasing. These reading habits help you more reliably detect AI in Substack newsletters when the detector’s score is unclear.
Q: Can writers disclose their use of AI on Substack and how should they do it?
A: Yes, writers can add an optional AI author’s note or a “how I make this” statement that explains which tools they used and where AI helped. Substack says the feature is meant to encourage transparent disclosures and stronger editing, not to prohibit or penalize AI-assisted writing.
Q: Are there limits to the Pangram detector’s accuracy or scope?
A: The detector estimates but does not prove authorship and can be less reliable on short or formulaic text, so it is best used with human judgment. The feature only scans items above 100 characters and Substack allows creators to report and remove scans they believe are incorrect.
Q: What can publishers do if Pangram flags their own content incorrectly?
A: Publishers can run Pangram on drafts before publishing to catch issues and can report or remove scans on published work if they believe a result is mistaken. Revising flagged passages, adding clear disclosures, and checking facts and links are practical steps to correct mislabels.
Q: Will Substack penalize or reduce visibility for writers who use AI?
A: Substack clarified the tool is not intended to prohibit or penalize AI-assisted writing and is designed to encourage honest process statements and better editing habits. The company says the feature aims to help readers trust content and to reduce “AI slop” in feeds over time.
Contents
