Vanta Pillar Security Agreement for AI Governance

Vanta has signed a definitive agreement to acquire Pillar Security, the two companies said on 9 November 2026. The Vanta announcement and a Pillar blog post by Dor Sarig, co-founder and chief product officer, and Ziv Karliner describe a signed deal, not a completed one. Neither post gives a price, a closing date or any remaining conditions.

The planned product is more specific than the deal terms. Vanta says a paid offering called AI Governance by Vanta is due in January 2027, with Pillar’s tools folded into that path over the first half of the year. Until the transaction closes and those features ship, the announcement is a timetable and a statement of intent, not a product customers can buy today.

The Vanta Pillar Security timetable runs into mid-2027

Vanta, which sells automated compliance and trust-management software, says it will connect Pillar’s runtime protection and adversarial testing to its Trust Graph, which it describes as a live map of a customer’s trust programme powered by over 400 integrations. In the account it gives, an agent that tries an unauthorised action would be blocked, after which a Vanta Agent would alert owners and write an audit trail.

That sequence is written in the future tense. Vanta sets out four staged capabilities once Pillar’s tools are integrated:

AI Inventory, in January 2027, to map AI applications, agents, models and MCP servers and connect them to owners, approved uses, data access, vendors, risks and controls.
Runtime monitoring and detection, also in January 2027, to spot suspicious activity and guardrail violations, with investigations prioritised using context from the Trust Graph.
Red teaming, in spring 2027, described as adversarial testing tailored to a customer’s business context.
Runtime enforcement, in summer 2027, to turn governance requirements into boundaries that block unauthorised actions before they happen.

The paid AI Governance by Vanta offering is scheduled for January 2027, in parallel with inventory and monitoring, and ahead of the later testing and enforcement stages. Those are Vanta’s target dates, not confirmation that the features have already been built, priced or released.

Vanta presents Pillar as a runtime security platform for AI applications and agents. Pillar, founded in 2023 according to its own post, says it discovers, tests and secures those systems across their lifecycle and blocks unauthorised actions at runtime. Sarig, quoted in the Vanta post, said: “An agent can cross a line in a single action, long after its last review. That’s why security has to live where agents act: test them the way attackers will, and stop them the moment they go out of bounds.”

Phil Harris, research director at IDC, is also quoted in Vanta’s announcement. He said “CISOs need controls that sit in the path and act in real time” and that “moves like this one show the market shifting from AI governance on paper to governance at runtime.” That is an analyst’s view, carried in the buyer’s own post.

Customer numbers and SAIL are the companies’ own claims

Vanta says “over 20,000 customers rely on Vanta to earn and prove trust.” Pillar’s post uses a slightly different wording, calling Vanta “the leading Agentic Trust Platform, helping more than 20,000 companies earn and prove trust.” Both figures come from the companies; neither post breaks them down by product, region or paying status.

Pillar also describes SAIL, a framework it created for securing AI agents. Vanta says the framework is aligned with ISO and NIST standards and OWASP guidance. Pillar says SAIL “has passed 50,000 downloads, and maps 91 AI risks across seven lifecycle phases.” Those download and risk-map figures are vendor claims. A download count is not the same thing as paying customers, nor evidence that the 91 listed risks are complete or independently assessed.

Pillar’s post also says its research has uncovered critical vulnerabilities rated CVSS 10, including a sandbox escape in n8n and a flaw in Google’s Gemini CLI. That is Pillar’s account of its earlier work, offered as background for the deal, not a new, separately documented disclosure in these announcements.

A signed agreement still leaves the hard part ahead

A definitive agreement is a real legal step, and both companies have now said the same thing in public. It is still not the same as a closed acquisition. There is no purchase price in the official posts, no date by which the deal is expected to complete, and no statement of what would happen to Pillar’s existing products or customers if the close were delayed or abandoned. Readers should treat the January-to-summer 2027 dates as a plan attached to a deal that has not yet finished.

The commercial pitch is clear enough, and it is also unproven on the evidence published this week. Vanta argues that governance, risk and compliance systems hold business context that can drift from what agents actually do, while point AI-security tools see threats without that context. Joining the two, it says, would turn policy into enforceable runtime boundaries. That is a coherent product thesis. It will only be testable once inventory, monitoring, red teaming and enforcement actually appear in the paid offering, on the dates Vanta has given, and once customers can see what is blocked in practice rather than in a blog example.

Other vendors are already selling pieces of the same problem. CrowdStrike recently paired an AI kill-switch with deals for Fiddler and Dreadnode, another case of a security platform buying AI monitoring and offensive-testing capability rather than building all of it in-house. ServiceNow has announced AI Passport, a control layer meant to limit what agents and MCP servers may do at runtime, which is close to the enforcement end of Vanta’s timetable. Microsoft’s preview of Entra Agent ID lifecycle controls is aimed at a different layer again: giving each agent an identity and a life cycle, which is the sort of ownership and inventory problem Vanta says it will tackle in January. None of those products proves that Vanta’s combination will work. They do show that the gap Vanta is describing is one other large firms are also trying to close.

The next facts that would change the story are straightforward. A close announcement would turn the agreement into an acquisition. A disclosed price would let readers judge the scale of the bet. And a January 2027 release, with clear packaging for the paid AI Governance offering, would show whether the first half of the timetable survived contact with a real product. Until then, Vanta Pillar Security is a signed deal and a dated roadmap, not a finished merger and not a shipping control plane.