Substack is giving readers a way to check whether the post they’re reading was written by a human or by a chatbot. The company has partnered with AI-detection firm Pangram to introduce new tools that will let users scan posts, notes, and replies for AI-generated text. CEO Chris Best introduced the features in a post titled “Against Claudefishing,” his term for content that leans on AI while presenting itself as human work.
How the scanning tool works
Scans run on any text longer than one hundred words that is published from July 21 onward, and the result is visible only to the person who requests it. Best says Pangram is not perfect, but points to independent evaluations that credit it with a high degree of accuracy. The feature is live on web and iOS, with an Android version coming later.
Substack’s AI-detection feature arrives at a time when the problem it targets is spreading fast across social platforms. Best cites a recent Pangram estimate that as much as forty percent of posts on some platforms are now fully AI-generated. This figure underscores the growing challenge of distinguishing human-authored content from machine-generated text, a concern that has intensified since the public release of large language models like ChatGPT in late 2022.
Writers can run the same scan before they publish
Readers are not the only ones getting new controls. Substack is adding a “How I make this” statement that lets writers describe their process and set expectations up front. Writers also get the option to run Pangram on their drafts before publishing, and they can report and remove scans of their published work that they believe are mistaken. Best says future additions could include AI preferences inside Reply Rules and reader-side controls for what gets recommended.
For readers, this is a rare transparency feature that hands them a real button to spot AI slop. Substack is not alone in trying to solve this issue. LinkedIn has started reducing the reach of AI-generated posts and comments, and Meta is quietly testing its own AI detection tool, though it has yet to go live. Whether any of these tools can keep pace with the models they’re built to catch remains to be seen.
The rise of AI-generated content has become a major topic of debate in the publishing industry. Platforms like Substack, which rely on individual creators and newsletters, face unique challenges. Readers subscribe to newsletters expecting a personal voice and authentic perspectives. If that voice can be replicated by an algorithm, the value proposition of the platform diminishes. By introducing detection tools, Substack aims to preserve trust between writers and their audiences.
Pangram, the company behind the detection technology, claims high accuracy in identifying text produced by large language models. However, no detection method is foolproof. False positives can occur, potentially flagging human-written content as AI-generated. To address this, Substack allows writers to dispute results and remove scans from their work. This appeals process is crucial for maintaining fairness and avoiding unjust penalties against creators who write in a clear, structured style that might resemble AI output.
The broader context of this move involves the entire ecosystem of online content. As AI writing tools become more sophisticated, detection must evolve in parallel. Some critics argue that detection tools create an arms race, with AI models being trained to evade detection. Pangram’s approach reportedly focuses on identifying statistical patterns common to machine-generated text, but adversarial techniques can sometimes circumvent these methods.
From a writer’s perspective, the “How I make this” statement offers a proactive way to signal transparency. Writers can describe whether they use AI for research, drafting, editing, or not at all. This allows readers to make informed decisions about the content they consume. It also reduces the stigma around legitimate AI usage, as many writers employ AI tools for brainstorming or grammar correction without replacing their own creative work.
Substack’s move could influence other platforms to adopt similar measures. If readers begin to expect transparency about AI use, platforms without such features may lose credibility. However, implementation costs and technical challenges remain barriers. Small platforms may lack the resources to develop or license detection tools, while large platforms worry about false positives affecting user experience.
The timing of Substack’s announcement aligns with growing regulatory interest in AI-generated content. The European Union’s AI Act, for example, includes provisions for transparency labeling. While Substack’s tool is voluntary and reader-initiated, it positions the company ahead of potential legal requirements. It also provides a model for how platforms can balance innovation with ethical considerations.
In practice, the detection tool is simple to use. On the web or iOS app, readers can click a button to run a scan on any eligible post. The result appears as a percentage likelihood of AI generation. Substack does not automatically label posts; the scan is performed only at the reader’s request. This respects both the writer’s privacy and the reader’s curiosity, avoiding blanket judgments.
For writers, the pre-publication scan is equally straightforward. Drafts can be checked in the editor, giving writers a chance to review their work before sharing. If the tool flags content the writer believes is human-written, they can adjust or ignore the result. Over time, this feedback loop could help writers understand how their writing style compares to AI patterns, potentially improving their craft.
Critics point out that the tool only applies to text published after July 21, 2025, leaving older posts unscanned. This limitation means that a large archive of past content remains unverified. Substack has not announced plans to retroactively scan older posts, likely due to the computational cost and the fact that many older posts were written before modern AI writing tools existed.
Another limitation is language support. Pangram’s models are primarily trained on English text, so scans may be less accurate for posts in other languages. Substack hosts newsletters in many languages, and the detection tool may not be equally reliable for all of them. The company has not commented on plans to expand language coverage.
Despite these limitations, the introduction of AI detection on Substack marks a significant step toward addressing the problem of AI-generated content in a transparent and user-empowering way. It gives readers a tool they can choose to use, rather than imposing a system that might penalize writers unfairly. It also gives writers control over how their work is perceived, allowing them to defend their originality if false flags occur.
The success of the feature will depend on its adoption and accuracy. If readers find it useful, they may start to expect similar features elsewhere. If false flags become common, trust in the tool could erode. Substack and Pangram will need to continually refine the detection algorithm and respond to feedback from both readers and writers.
In the competitive landscape of newsletter platforms, transparency could become a differentiator. Substack already faces competition from platforms like Ghost, Beehiiv, and Revue (owned by X). Adding AI detection could attract writers who want to signal authenticity to their subscribers. It also aligns with the platform’s emphasis on independent, human-driven publishing.
The concept of “Claudefishing,” as Best called it, refers to content that presents itself as human but is actually generated by an AI model like Anthropic’s Claude. The term plays on “catfishing,” where people pretend to be someone else online. Best argues that readers deserve to know if what they’re reading is genuinely crafted by a person or produced by a machine.
As AI continues to permeate content creation, the line between human and machine writing will blur further. Substack’s proactive stance may not be perfect, but it opens a conversation that the industry cannot afford to ignore. By giving both readers and writers tools to navigate this new reality, Substack is taking a practical first step toward a more transparent internet.
Source:Digital Trends News