ZeroClaw is a technically impressive, security-conscious self-hosted AI agent runtime, but the absence of a legal entity, no published ToS or Privacy Policy on the official website, and an active impersonation ecosystem mean you need to verify your source before deploying.
ZeroClaw is dual-licensed MIT/Apache-2.0, fully open source, and self-hostable — all data stays on the user's own hardware by design; no cloud telemetry was found in documentation (source: https://github.com/zeroclaw-labs/zeroclaw).
The official website at zeroclawlabs.ai carries no published Terms of Service or Privacy Policy page; the only governance document available is the open-source LICENSE files and SECURITY.md (source: fetch of https://www.zeroclawlabs.ai/).
The project team publicly disclosed that using Claude Code OAuth tokens outside of Claude Code/Claude.ai violates Anthropic's Consumer Terms of Service — the README warns users to avoid this integration path (source: https://github.com/zeroclaw-labs/zeroclaw README announcements table).
Active impersonation is a documented risk: zeroclaw.org, zeroclaw.net, and the openagen/zeroclaw fork are confirmed as unauthorised by the official team; users must verify they are using the canonical github.com/zeroclaw-labs/zeroclaw repository (source: https://github.com/zeroclaw-labs/zeroclaw README announcement, Feb 2026).
No CVEs or public vulnerability disclosures were found for ZeroClaw on NVD/MITRE; the project has 129 automated security tests, a SECURITY.md with a 48-hour acknowledgement SLA, and CIS Docker Benchmark compliance enforced in CI (source: https://github.com/zeroclaw-labs/zeroclaw/blob/master/SECURITY.md).