Anthropic wanted assurances that its AI would not be used for fully autonomous weapons or for domestic surveillance.