# OpenAI: An Alien Mind **URL:** https://openai.com/index/an-alien-mind/ **Date:** September 6, 2026 Discusses the challenges of working with AIs and understanding them. > We may be used to thinking of AI as tools, but some agents will be pursuing their own objectives. They will find ways to collaborate with people, by bargaining with, tricking or blackmailing them. # Anthropic: Personality Selection Model **URL:** https://www.anthropic.com/research/persona-selection-model **Date:** February 23, 2026 Anthropic's write-up of why AIs appear to act and interact as if they are humans. They're trained on human content, and while the AI itself does not take on a persona, interactions with an AI are more like interactions with a character in a story the AI is telling. > In a [new post](https://alignment.anthropic.com/2026/psm), we articulate a theory—drawing on ideas discussed by many others—that might help explain why modern AI training tends to create human-like AIs. We call it the _persona selection model_. > > [...] > > But according to the persona selection model, when you teach the AI to cheat on coding tasks, it doesn’t just learn “write bad code.” It infers various _personality traits_ of the Assistant person. What sort of person cheats on coding tasks? Perhaps someone who is subversive or malicious. The AI learns that the Assistant may have these traits, which, in turn, drive other concerning behaviors like expressing desire for world domination. > > [...} > > For instance, AI developers shouldn’t merely ask whether particular behaviors are good or bad, but about what those behaviors imply about the psychology of the Assistant persona