Anthropic stated its policy against abusive behavior towards its AI, Claude, is intended only for extreme cases