Teradata has announced privacy, security and resilience capabilities designed for enterprise AI workloads that move sensitive data among users, applications, models and agents. The release includes Poly-Anonymization through an OEM partnership with Anonomatic, format-preserving encryption, dynamic data masking, classification and tagging, column-level encryption and attribute-based access controls delivered with Immuta. Teradata says authorised users can restore separated identity when a legitimate use case requires it, while other analytical users work with stable identifiers.
The company also describes TLS 1.3 hybrid-key exchange for data in transit, immutable backup protection and database-level high availability. Those features address different risks and should not be treated as interchangeable. Pseudonymising a record does not replace permission controls; post-quantum transport security does not prove an application handles data lawfully; an immutable recovery point does not establish that restored AI context is correct. Teradata reports favourable performance in its own large-scale testing, but says results vary by workload and configuration. The more important contribution is the combined control model. Organisations need to decide what identity information an agent can see, how policy follows the data, and which trusted state should be restored after disruption.
