Gather the right information
Choose the AI events and context your application will record.
Loading...
Use cases
Record the identity and workflow behind an AI action so another person can examine its source and history.
AI records
A model response may sound plausible while its supporting record is missing or altered. Verify the evidence attached to the output, and keep factual assessment separate from the cryptographic checks on that evidence.
Truthlocks does not determine whether arbitrary online content was created by AI merely by inspecting it. Traceability requires recorded evidence.
Use cases
Choose the AI events and context your application will record.
Check the signatures and inspect the information supplied by the connected system.
Share the record with reviewers and connect follow-up actions to the same process.
Details and limitations
Choose the point in your application where a reported AI action becomes a signed record. Configure the integration in Console, including its identity and environment. Capture enough information to interpret the action without copying private prompts or confidential output into a public receipt. A record can describe supplied context; it cannot reveal events the integration never captured.
Generate a non-sensitive summary, retain the evidence reference and inspect its receipt. Alter a test copy of a signed field and check the verifier's result. Confirm that the interface explains what was signed and which claims remain unverified rather than treating a cryptographic signature as a quality score for the AI output.