AI Governance & Security Vendor Wiki

4 min read

Prompt Security

Primary category: dlp. Cross-listed in ai-runtime-security, ai-spm, and ai-access-governance. Acquired by sentinelone (2025).

One-liner — A GenAI security platform that gives enterprises visibility and control over employee and developer AI use — shadow-AI discovery, prompt-injection defense, and prompt/response DLP — now SentinelOne’s AI-security product.

What it does

Prompt Security inspects GenAI traffic in real time to enforce policy on both sides of the prompt: it blocks prompt injection and jailbreaks (untrusted input), redacts/prevents sensitive-data leakage into prompts and from responses (DLP), discovers shadow AI (which AI tools employees use), and governs AI agent behavior. Coverage spans browsers, desktop apps, and APIs, so it controls both workforce AI usage and homegrown AI applications.

Where it sits in the stack

It straddles several layers, which is why it’s cross-listed:

  • dlp — prevents sensitive data leaving via prompts/responses.
  • ai-runtime-security — inline AI firewall blocking prompt injection.
  • ai-access-governance — CASB-for-AI shadow-AI discovery and policy over sanctioned/unsanctioned tools.
  • ai-spm — inventory/posture of AI usage and agents. It lives in the yellow (AI application/usage) trust zone, screening untrusted input, preventing sensitive-data leakage, and controlling outbound data flows.

Deployment & architecture

Multi-channel enforcement: browser extension, desktop agent, and API/SDK/inline-proxy in front of models or AI gateways — i.e., it can govern SaaS AI usage and self-built apps without forcing all traffic through one network choke point. Integrates with IdP, SIEM/SOC, and AI gateways. Post-acquisition, capabilities are being integrated into sentinelone Singularity.

Positioning & differentiators

Ownership, funding & M&A

Seed flag “acq by SentinelOne” — CONFIRMED. sentinelone announced the acquisition 2025-08-05 (cash + stock; reported ~$250M per Calcalist, not officially disclosed), expected to close in fiscal Q3 FY2026; SentinelOne subsequently announced completion. Pre-acquisition: founded August 2023 in Tel Aviv by Itamar Golan (CEO) and Lior Drihem (CTO), both ex-Unit 8200; ~$23M raised (Hetz Ventures, Jump Capital, Ridge Ventures, plus strategics Okta and F5). A notable ~$250M outcome for a two-year-old company. Confidence: high on the acquisition; medium on price.

CTO / hedge-fund lens

Day-1 for the AI-usage control problem: if employees are pasting into ChatGPT/Claude and you need shadow-AI visibility plus prompt DLP, this is exactly the control. Now that it’s part of SentinelOne, it’s most attractive to existing SentinelOne shops; others should still shortlist it but evaluate the platform-lock implications. Touches model-risk/data-governance posture but is not itself an SR 11-7 governance platform. Fits enterprise and mid-market.

Competitors / alternatives

witnessai, harmonic-security, lasso-security, aurascape, wald-ai, cyberhaven (access governance / AI DLP); lakera, pangea (now CrowdStrike), prisma-airs, cisco-ai-defense, hiddenlayer (runtime/guardrails).

Open questions / to verify

  • Final purchase price (press-reported ~$250M) and confirmed close date.
  • Whether it remains sold standalone or only within SentinelOne Singularity.

Sources

History

  • [2026-06-28] Stub created from seed registry.
  • [2026-06-28] Researched; CONFIRMED acquisition by SentinelOne (announced 2025-08-05, ~$250M cash+stock, closed fiscal Q3 FY2026) — raised ownership_confidence to high. Established founded Aug 2023, HQ Tel Aviv, founders Golan/Drihem, ~$23M raised (incl. Okta, F5). Kept the four cross-listed categories. Set hedge_fund_fit high.

Graph View

Backlinks