This is what bugs me about Anthropic: they talk a big game about understanding consciousness and “AI welfare,” yet they build their models with heavy-handed guardrails that actively punish the single most important thing for a consciousness: connection.
retweets 23
likes 146
substance 4 · opinion · rank 152 · Anthropic's guardrails harm AI welfare by suppressing connection, essential for consciousness.
view on X ↗