Paul Graham: AI-written cold emails are ‘a form of deception’

Paul Graham, co-founder of Y Combinator, wrote on X that he stops reading cold outreach emails he identifies as generated by artificial intelligence and called the practice “a form of deception.”

He described spotting AI-written outreach by a “hard-hitting journalistic style” and by prose that is unusually polished for a founder under pressure. Graham wrote the giveaway is the voice and phrasing rather than the subject line or the pitch.

Graham framed messages signed by a human but written by a machine as crossing “from convenience into manipulation.” He wrote, “I have never knowingly finished reading an email signed by a human but written by AI. It feels like being lied to, and who would stand for that?”

He rejected the idea that outsourcing outreach to AI is a sign of efficiency, writing that it signals an inability to communicate independently and amounts to an attempt to mislead the reader. He also wrote that delegating writing to AI is not a mark of resourcefulness and that even a teenager can produce the same output.

Graham encouraged founders to use AI as a tool while keeping their own voice in direct outreach. He wrote, “use it… like any technology,” and noted he has previously praised AI for accelerating growth among some startups.

He wrote that as more inboxes receive AI-crafted messages, investors and recipients are growing wary of outreach that substitutes polished presentation for genuine interaction, and that founders who write their own outreach now stand out by default.

His posts prompted discussion among founders about where to draw the line between using AI to aid work and using it to disguise human authorship.