Mountain View, California, November 5, 2026 / PRNewswire / -- Cyberhaven, which provides data security services to "agency-based enterprises" ( agentic enterprise ), today launches Agentic Data Security ( ADS ). The company states that this product will help chief information security officers ( CISO ), data security teams, internal risk teams, and AI governance officials to say "yes" to AI agencies without losing control over the data.

Cyberhaven indicates that Agentic Data Security allows security teams to see how each AI proxy handles company data, on behalf of whom it acts, and where the data goes next, allowing them to stop it before sensitive information is leaked. The company claims that from cutting-edge AI laboratories to top-tier quantitative hedge funds, those enterprises that are most proactive in embracing a proxy-based future have already relied on Cyberhaven to protect their most sensitive data. ADS is now available for purchase.

The company stated that AI agents have changed the way work is completed, and at the same time, they have introduced new internal risks. As these agents transition from pilot projects to production environments, security officials in various industries are raising the same question: How are these agents handling our data, and who is initiating these operations? The company explained that agents operate with the access permissions of the people they serve and perform tasks at machine speed; therefore, a single agent can read files, query databases, and share what it finds within a single workflow. Cyberhaven Labs observed that the adoption rate of terminal-based AI agents has doubled year-over-year, reaching 60% by May 2026. The company noted that this rapid adoption has brought about new data risks, which may originate from "trusted" agents.

Cyberhaven CEO Nishant Doshi indicates:"AI proxies now operate with the access permissions of trusted employees. For most companies, the first proxy-related data incident might be when a proxy acts exactly as required but accesses data that it shouldn't have. Companies tend to adopt proxies faster than they establish new control measures for them. Agentic Data Security allows them to deploy proxies with more confidence while maintaining control over the data itself, rather than just over the tools that access the data."

Agentic Data Security is the latest module within Cyberhaven Flow. Cyberhaven Flow is a unified data security platform that utilizes data lineage (data lineage), AI, and autonomous protection to provide data loss prevention (DLP), internal risk management (IRM), and data security posture management (DSPM) for both humans and agents. The company stated that this also represents the next step in its technical strategy proposed in July of this year, which combines data lineage with identity and behavior to protect data in every human and agent workflow.

Industry analysis firm SACR Research Director Lawrence Pingree stated:It is impossible to manage the risks of the AI era without understanding which data is accessed, by whom it is accessed, through what workflows it is accessed, and whether these accesses are authorized. The data from Cyberhaven provides a foundation for security teams to cover both personnel and AI agents on the same platform simultaneously.

Retain proxies while also maintaining control over the data.

Cyberhaven indicates that its Agentic Data Security provides four controls for each agent:

• Discover every proxy. Identify each AI tool and proxy running in terminals, any cloud environment, or browsers, including unauthorized applications, and assign a risk score to each of them.
• Know who sent what. Link each action with the person or system behind it, and view both the personnel and agents in the same view at the same time.
• Stop it before the data leaves. Apply existing policies designed for personnel to proxy operations in order to identify high-risk behaviors and take control before sensitive data is released, while not hindering the team from adopting AI.

Datavant Chief Information Security Officer Dan Walsh states:Proxy-based enterprises will be defined by the company's control over the data that proxies come into contact with, rather than by the number of proxies they deploy. Proxies have changed the way data leaks are calculated: they move faster, on a larger scale, and have trusted access rights. The winners will be those companies that can say 'okay' to AI without having to watch data flow out, and this requires a security capability that can follow the data wherever it goes. This is precisely the direction that Cyberhaven is ready to lead.

Cyberhaven indicates that its technology can track the entire lifecycle of data and adjust protection methods according to changes in context. Agentic Data Security brings this capability to agents: by implementing data lineage, it provides full visibility into workflows, covering both humans and agents with a unified architecture, and takes protective actions in real-time through execution mechanisms across terminals, browsers, and cloud environments. The company claims that ADS applies data lineage and a unified policy engine to agents without the need to add new tools, consoles, or sets of policies.

Cyberhaven Chief Product Officer Manish Kalia states:The data security of the proxy era will be defined by our understanding of how to handle AI data. Our proprietary AI model helps customers not only classify their data accurately but also understand the flow of data during mobileization and transformation processes. As the call and reasoning trajectories of proxy tools are precisely what Cyberhaven is already capable of checking, the strategies that customers use to manage internal risks can now also be applied to proxies.

Learn more about how to protect agency-based businesses:

• Attend the executive live briefing held at 11:30 a.m. Central Time on December 3rd: “Double Trouble : Taking Control of Human & Agentic Insider Risk”.
• Read the blog post from Manish Kalia: `Agentic Data Security : Control What AI Apps and Agents Do With Data`.
• Read the October 2026 article from MIT Technology Review Insights: Securing the agentic enterprise with data lineage.
• Watch the demonstration video for Agentic Data Security: https :// vimeo.com /1235907613? fl = pl & fe = sh

About Cyberhaven

Cyberhaven indicates that the company is a leader in the field of data security for agency-based enterprises, trusted by the world's most innovative companies to protect their most sensitive data. Cyberhaven enables tracking of the entire lifecycle of data and adjusts protection methods as context changes. The company claims that it protects not just the data itself, but also the workflows; by connecting data lineage, identities, and behaviors, it understands how data is created, copied, divided, and shared within these workflows. Cyberhaven operates in real-time scenarios where personnel and agencies process data, helping to prevent risks from evolving into losses. The company states: “AI has changed the way work is done, and Cyberhaven protects this new method of working.”