Teradata's new security measures aim to harness AI agents
The vendor's new features, including poly-anonymization to reduce PII exposure address customer needs, bring its privacy and security capabilities in line with its peers.
Teradata is taking a stab at taming the security challenges created by AI agents, unveiling new privacy and resilience capabilities to protect sensitive data.
To combat the emerging security risks posed by agents, Teradata on Tuesday introduced new capabilities, including poly-annoymity to reduce identity exposure while preserving the relationships between data points that AI requires; format-preserving encryption including Anonomatic's PII Vault to protect sensitive values without changing their format; and post-quantum security and readiness to improve protection against emerging threats.
"The new capabilities represent meaningful operational upgrades for existing customers," Donald Farmer, founder and principal of TreeHive Strategy, told TechTarget.
However, he noted that Teradata is catching up to the AI security measures some of its competitors already have in place rather than introducing innovative capabilities that will force the rest of the market to react.
"These features only bring Teradata up to par with modern cloud data platforms rather than establishing a technical moat," Farmer said, naming AWS and Cloudflare among those already offering some of the tools Teradata is adding. "[But] to be fair, Teradata is among the earlier relational database vendors to market post-quantum transit encryption and PII vaulting."
Addressing need in the age of agents
While Teradata's new tools bring the vendor's security and privacy protection are not new to the market, they are nevertheless critical for its customers as they increasingly put autonomous AI agents into production.
Customers want to use this data for AI, testing and collaboration, yet exposing raw identity creates regulatory, security and trust barriers. We built toward protecting the data while preserving its relationships and usefulness.
Sumeet AroraChief product officer, Teradata
It's those customers whose struggles protecting their information pushed the vendor to advance its AI security measures, according to Sumeet Arora, Teradata's chief product officer. He noted that as AI draws sensitive, individual level data into its workflows, the masking methods that were enough when only humans needed to be controlled are no longer sufficient.
"Customers want to use this data for AI, testing and collaboration, yet exposing raw identity creates regulatory, security and trust barriers," Arora said. "We built toward protecting the data while preserving its relationships and usefulness."
In addition to poly-anonymization, format-preserving encryption and post-quantum security and readiness, Teradata also introduced immutable backup protection to strengthen cybersecurity readiness and a new database availability layer for mission-critical workloads that is different from disaster recovery.
Privacy capabilities provided by Anonomatic and delivered by Teradata are generally available, while data security and protection capabilities such as poly-anonymization, format-preserving encryption, post-quantum and immutable backup protection are available to customers based on the Teradata platform they use and their deployment type.
"These enhancements are right on target with enterprise priorities," Kevin Petrie and analyst at BARC U.S., told TechTarget. "Our research shows that privacy is by far the number one priority of responsible AI, followed by security."
However, like Farmer, he noted that Teradata, while serving the needs of its existing customers, is adding features that certain competing vendors already provide rather than building innovative new features that will force others to respond.
"Other vendors such as IBM and Oracle also have encryption capabilities that address quantum-era threats," Petrie said.
More to do to manage agents
As agents become more ubiquitous, Arora noted that the data landscape is becoming more complex. Unlike BI and machine learning applications that generally operate in a single system, agents operate across cloud, on premises, hybrid and sovereign environments.
Teradata is therefore focused on building tools that address customer concerns as they expand their agentic AI networks.
"We are focused on creating solutions for the enterprise concerns that will expand as this happens, such as context, governance, cost-optimization, speed and generally supporting the new architecture that agentic AI requires," Arora said.
Farmer suggested that while the security and privacy controls Teradata is adding are valuable, the vendor would be wise to add even more. Data governance and security specialist Immuta is providing tools such as Attribute-Based Access Control that determines what user can access data at the exact moment a query runs rather than use role-based access control to check if a user is part of a preapproved group.
A similar tool would improve Teradata's AI security suite.
"Teradata should look at Immuta's work," Farmer said. "Immuta’s platform grants, denies or masks data based on the intersection of user behavior, data classifications and context. Rather than managing hundreds of unique database roles, data engineers use Immuta to build plain-English rules that scale automatically across cloud platforms. ... This feels like the next thing Teradata should be doing."
Eric Avidon is a senior news writer for Informa TechTarget and a journalist with more than three decades of experience. He covers analytics and data management.
