Zscaler Targets AI Identity Risk With Symmetry Acquisition

Agentic AI , Artificial Intelligence & Machine Learning , Identity & Access Management Zscaler Targets AI Identity Risk With Symmetry Acquisition Startup Symmetry Systems Maps Relationships Across AI, SaaS and Cloud Assets Michael Novinson (MichaelNovinson) • May 22, 2026     Share Post Share Credit Eligible Get Permission Dhawal Sharma, EVP, AI Security and Strategic Initiatives, Zscaler (Image: Zscaler) Zscaler plans to purchase an access graph startup founded by a University of Texas professor to understand the lineage of artificial intelligence systems, identities, applications and data. See Also: Know Thy Enemy: Threats to Cyber Resilience The San Jose, California-based cloud security titan said its proposed acquisition of San Francisco-based Symmetry Systems will give Zscaler visibility into where AI is being used, what data it accesses and how those systems interact with identities and applications, said Dhawal Sharma, executive vice president of AI security and strategic initiatives. He said AI visibility is often spread across disconnected security products. "AI is essentially extending its tentacles everywhere from how employees are doing day-to-day work on internet sites," Sharma told ISMG. "Bringing all this information together is very complicated." Symmetry Systems, founded in 2019, employs between 40 and 45 people and raised $35.7 million in capital, having most recently completed a $17.7 million Series B funding round in August 2023 led by Prefix Capital. The company has been led since its inception by Mohit Tiwari, who has been a professor at Texas for the past 13 years focused on security, privacy and computer architecture research (see: Zero Trust Anchors AI Security Strategy). How Symmetry's Technology Can Map Relationships Zscaler operates endpoint agents, monitors inline traffic and scans public cloud environments, but firms struggle to connect those separate datasets into a cohesive understanding of AI risk, Sharma said. Symmetry's access graph technology unifies disparate systems and creates contextual relationships between data, identities, SaaS applications and AI assets, according to Sharma, he said. "The bigger problem becomes, 'How do you connect the dots of these various different systems on how you build everything that connects to AI and understands its usage, its lineage and associated risk?'" Sharma said. "This is where Symmetry's access graph technology comes in very handy, because they already been doing this." Symmetry's technology helps organizations map relationships among identities, SaaS applications, datasets, AI models and cloud resources in a highly contextualized way, Sharma said. The firm clusters relationships into smaller contextualized environments tied to identities, applications or datasets, allowing clients to detect anomalies more effectively and turn graph data into actionable intelligence. "What Symmetry brings uniquely with Zscaler is being able to say we have that data access figured out for who's accessing what, but also the context of AI applications, which models and pipelines it is going to," Sharma said. "So, it gets more contextualized, and being able to build more fine-grained policies on top of it." Organizations now need visibility into how identities transform as AI systems interact with applications, APIs and data pipelines, which requires the need to understand complicated IAM structures across hyperscalers and SaaS platforms. Symmetry's technology provides deep visibility into identity relationships within SaaS environments and public cloud infrastructure, enabling much more granular policy enforcement. "The non-human identity, agentic identity is a harder problem because the access provided to these entities is either based on a human identity, or it is tied to a service principle if it is an autonomous agent," Sharma said. "There's nothing in between on which you can build fine-grained authorization." How Traditional Security Operations Will Fare in an AI World AI workflows constantly create new identities, service relationships and permission structures as tasks move between models, pipelines and autonomous agents, and enterprises lack tools capable of tracking these transformations across multiple environments simultaneously, he said. Zscaler wants to combine identities, AI models, applications and data relationships into a single unified framework, Sharma said. "The problem we are solving is multi-channel, multi-model lineage of data and identity in full context of AI," Sharma said. "That is the problem which certain startups, vendors have solved for a channel or a given type, but not across the board. This is the hard problem that we are solving." Zscaler plans to deepen endpoint AI visibility and improve lineage tracking using Symmetry to better understand how AI applications connect to identities, datasets and infrastructure components, Sharma said. And combining identity-aware lineage mapping with Zscaler's zero trust architecture will help organizations build more granular controls around AI usage and data access for tools such as MCP. "We are taking a very platform-centric approach to find all the AI and everything that connects to AI, secure the usage of it and then govern it," Sharma said. "This build gives us that foundational layer at seeing everything in AI, and then being able to build pieces of technology that will govern what goes on in the AI as well." Symmetry-generated detections and lineage intelligence will become another high-value telemetry source, generating higher-fidelity alerts and reducing the false positives that often burden traditional security operations environments, Sharma said. Specifically, he said Symmetry's identity and lineage insights are expected to improve incident triage and remediation workflows significantly. "Our goal is to acquire companies that become features on the platform in longer run, and they basically solve the complex problem that we are trying to solve by integrating that into our platform," Sharma said.