AgentRecords   TrackerChangelogAbout / Sponsor

Claude (Anthropic)

Tiers: Consumer (Free/Pro/Max); Team/Enterprise; API

RetentionConsumer: deleted chats purged from back end within 30 days. If model-training is on, de-identified data kept up to 5 years. Flagged chats: inputs/outputs up to 2 years, safety scores up to 7 years. Feedback: 5 years
TrainingConsumer: a user setting (model improvement); default not stated on checked page. Commercial: see the org-data article (not reviewed in detail)
eDiscovery / exportNot stated
Admin controlsCommercial controls: not stated on checked page
Review statusverified-partial

Official sources (checked 2026-10-15)

Weekly "what changed" email
(placeholder, no data is collected yet)

Not legal advice. Summaries of public vendor documentation as of 2026-10-15. Always check the linked official source.