It’s getting harder to tell what’s real on the internet.
This is less true on Substack, and we aim to keep it that way. As a first step, we’re launching a feature with Pangram in the Substack app today. Going forward, you’ll be able to scan text to see how much of it was likely written by hand or with AI assistance. More on that in a moment, but first let me explain why this matters.
Water, water, everywhere, nor any drop to drink
There is an increasing share of text on social media that is generated by AI—as much as 40% on some platforms, according to Pangram’s estimate. This is a big change. Many people complain their feeds are filled with soulless slop. Others don’t yet realize when they are reading something written by no one.
The core problem is not people using AI, or the quality of its output. Not everything made with AI is slop, and not all slop is made with AI. The problem is when there is a mismatch between a reader’s expectation and reality, especially when they unwittingly invest their attention in something with no human thought on the other end. That’s Claudefishing.
Here’s how Freddie deBoer put it:
I access human-made art because I know there’s a human behind it and that’s what I’m looking for, other humans, showing me in art what they hide in their selves. Fooling me in that process is just a con.
We care about this at Substack because it gets to the core of what we do: building an economic engine for culture. That means helping real people, who think, feel, observe, argue, experiment, consider, struggle, exhort, and care, to make money doing work they believe in. The network is based on trust between people, and that’s why it works.
When content made by no one takes over parts of the internet that are supposed to be human, it pollutes the commons and makes it hard to discover and hear human voices. When readers have to wonder if what they’re reading is real, it undermines trust in authorship and threatens the livelihood of writers—including those who use AI tools thoughtfully to produce work they believe in. Platforms that reward fakeness will create a race to the bottom.
What we’re doing about it
This is an emerging space, and we don’t claim to have all the answers. But one thing we do know is that we don’t want to wait until your Substack app turns into LinkedIn before we start to learn and make progress. So to start, we’re launching some new tools that we hope will increase transparency and give us a basis to ask for your feedback.
We’re partnering with Pangram, the leading AI-detection tool. You’ll be able to scan notes, replies, comments, and posts to see an estimate of how much of the text was written by hand or with AI assistance. This will work on text longer than 100 words, published from today on, and will show an analysis only to those who request it.
Note that Pangram can only detect whether AI was used to make the text, not whether great human care went into creating it, nor whether AI tools were used as a source. It is not perfect, but independent research suggests that it detects AI-generated text with a high degree of accuracy. We believe it is beneficial for people to have access to tools like this to help make their own judgments.
We’re also introducing tools for creators, including the ability to add a “How I make this” statement where you can explain your process and set expectations for readers. You can run Pangram on your drafts prior to publication, as well as report and remove any scans on your own work that you believe are mistaken.
These features are available now on web and iOS and will be coming to Android soon. This is just a starting point. Depending on feedback and interest, we are considering tools that will:
Let you set preferences around AI content for your own community in your Reply Rules
Give creators more ways to express the individual value they offer, whether through human writing or thoughtful AI use
Give readers the option to set preferences about what gets recommended to them
Continue to improve our systems that fight spam, bots, and scams of all kinds as AI accelerates these problems
We’re not against people using AI to assist their work, and we think people should be free to choose which tools they use to express themselves. Some on the platform are human-writing maximalists, others are publicly exploring the frontier of using AI tools while sweating the details to make work they stand behind. We use AI all the time at Substack to write software, do research, and build product features like clipping, translations, and more.
But people should know what they’re getting. For example, you might be fine reading someone’s sharp analysis in which they leaned heavily on AI to present their findings but only want to spend time engaging with replies written by a human hand. Either way, you should have a choice in the matter.
What do you want to see?
We’re all entering this strange new world together, and we can think of no better way to figure out the new tools and new social norms we need than open debate. We want to hear from you: How can Substack help support valuable human creation in the age of AI? When is AI use Claudefishing? When is it great?
Human perspective is the essential ingredient of culture, and this will be true no matter how powerful AI becomes. In fact, the more you expect these tools to keep improving, the more important it is to understand the provenance of what you read. Claude has a lot to offer! But when I want Claude’s opinion, I’ll ask Claude.
Comments are open for this one. Let’s go!
Oh look. Another purity test.
In eighteen months of publishing on Substack, not one reader has ever accused me of producing AI slop or suggested there was no human being behind my writing.
I use AI. I also bring the stories, memories, opinions, arguments, humor, lived experience, and final judgment. My readers seem perfectly capable of recognizing that a real woman is present.
A detection score may tell you that a tool was involved. It cannot tell you who had the original thought, who shaped the argument, who rejected the bad language, who supplied the life, or who stands behind every word.
Transparency is one thing. Turning authorship into a purity test is another.
The presence of AI does not prove the absence of a human.