Paul Graham Warns Against AI-Generated Founder Emails

Paul Graham, cofounder of Y Combinator, posted on X that many emails he receives from founders are now written in a "hard-hitting journalistic style" that he recognizes as AI-generated, Business Insider reports. Business Insider and Yahoo Tech quote Graham saying, "It feels like being lied to," and that he generally stops reading once he identifies an email as produced by AI. Business Insider also notes a 2024 post in which Graham said he watches for the word "delve" as an AI tell. Business Insider reported Graham replied to a commenter, "You're supposed to use it, but in the right way."

Editorial analysis - technical context: Large language models tend to produce polished, consistent prose that can be identified by stylistic patterns such as formalized transitions and uniform sentence length. Observers and practitioners have repeatedly noted that model-generated text often exhibits those markers, which can make AI-written cold outreach appear more professional but less idiosyncratic than genuine, hurried human writing.

Public reporting frames Graham's remarks as part of a broader debate about authenticity signals in startup fundraising and outreach. Yahoo Tech describes his stance as treating AI-ghostwritten messages as dishonest rather than merely efficient. This debate matters because founders use AI widely for drafting and editing, and investors and gatekeepers are part of the audience that will judge the credibility of outreach.

For practitioners: monitor how investors and accelerators publicly discuss authenticity, and watch for emergent markers that audiences cite as AI tells. Observers may also track whether community norms shift toward valuing self-authored outreach as a differentiator or whether tooling evolves to add provenance metadata. Coverage so far includes reactions such as a supportive post from Google DeepMind researcher Nataniel Ruiz, as reported by Business Insider.