Added Admins can now run Paladin and other AI features on their own Anthropic API key, so usage, billing, and data retention fall under their own Anthropic account Improved Paladin now evaluates AI agent access requests against the agent’s purpose, ownership, access risk, credentials, sponsor, tier and purpose tags; these signals appear on the request page and in Slack Access requests filed by a service user on behalf of a user now read “Requested for via “ on the request page and in notifications, instead of separate fields Admins can now preview a reviewer’s campaign view directly from the R