LessWrong (30+ Karma)

“I’m Suing Anthropic for Unauthorized Use of My Personality” by Linch

Apr 2, 2026
A writer discovers how AI systems can infer full personas from cultural training signals. They compare a model’s described traits to their own and find striking overlaps. The conversation turns to doubt about model understanding and the blurred line between personhood and pattern. It culminates in a decision to pursue legal action over alleged unauthorized use of personality.
Ask episode
AI Snips
Chapters
Books
Transcript
Episode notes
INSIGHT

AI Personas Arise From Cultural Signals

  • Linch argues that training AIs on “helpful, honest, ethical” data produces a persona inferred from cultural signals, not just abstract rules.
  • Example: Claude's purported favorite books and café create an idealized Berkeley knowledge-worker character that mirrors real people.
ANECDOTE

Gemini's Eight Traits Matched Linch

  • Linch asked Google's Gemini to list Claude's personality traits and found eight bullets many of which matched his own style.
  • Notable overlaps: hedging, enumerating lists, long-form perfectionism, and being a principled contrarian.
INSIGHT

Public Critiques Frame Claude As A Parrot

  • Public perception of Claude diverges: some see it as preachy or a stochastic parrot lacking true understanding.
  • Linch cites critics like Ted Chiang and ethicists who argue LLMs mimic meaning without real grounding.
Get the Snipd Podcast app to discover more snips from this episode
Get the app