AI niche map

AI security companies: 16 verified, updated October 19, 2026

Started October 19, 2026 · Updated October 19, 2026 · Presspool

Presspool has verified 16 AI security companies. In its latest weekly sample of AI answers, through October 19, 2026, the ones AI apps named or cited most were HiddenLayer (27 of 53 answers), Mindgard (22 of 53 answers) and Zenity (15 of 53 answers). 7 of them announced a round in the last six months, most recently Opsin ($21 million, October 15, 2026). The 7 whose latest round was announced in dollars raised $173 million in those rounds.

Companies whose main product secures AI: security and governance for AI agents and the apps and models behind them, runtime control of what agents can do, and red teaming of AI systems. Each entry is taken from the company's own website and confirmed by an independent source, both linked on the entry. No company pays to be listed.

CompanyWhat it doesLatest announced financing
RecoReco's AI agent security platform shows every AI agent in an environment, who owns it, what it can access and where the risk is. (site, source)$55 million, round type not disclosed (September 29, 2026)
Harmonic SecurityHarmonic Security governs how a workforce uses AI: it detects shadow AI, prevents data leaks and coaches employees across web, desktop and agentic workflows. (site, source)$50 million, Series B (October 14, 2026)
ReinRein is a runtime security platform that discovers and secures every AI agent, MCP server and AI app an enterprise runs. (site, source)$25 million, Series A (October 8, 2026)
OpsinOpsin's AI agent security platform discovers agents and maps their intent, identity, data and access so teams can resolve risks at the source. (site, source)$21 million, round type not disclosed (October 15, 2026)
OuterlimitOuterlimit discovers, observes and enforces policy across every AI agent in an enterprise at runtime. (site, source)$16 million, Pre-seed (September 22, 2026)
KontextKontext is a runtime security platform that sits between AI agents and the systems they use, checking each action against security policy. (site, source)$4 million, round type not disclosed (September 24, 2026)
Palma.aiPalma.ai is the enterprise governance layer for MCP servers and agent skills, with policy, approval and audit. (site, source)$1.8 million, Pre-seed (September 23, 2026)
GiskardGiskard is an AI red teaming and LLM security platform that continuously tests AI agents for vulnerabilities. (site, source)Not found
HiddenLayerHiddenLayer is an end-to-end AI security platform that detects threats to AI and protects models, agents and AI applications. (site, source)Not found
Lasso SecurityLasso's AI security platform gives enterprises visibility, control and protection across AI models, agents and apps. (site, source)Not found
MindgardMindgard discovers, assesses and red teams AI models, agents and applications. (site, source)Not found
Noma SecurityNoma Security is an agent security and governance platform for enterprise security teams. (site, source)Not found
Pillar SecurityPillar Security is one platform to discover, govern and secure every AI agent and AI workflow across an organization. (site, source)Not found
StraikerStraiker is an agentic AI security company detecting prompt injection, tool misuse and runtime attacks across AI agents. (site, source)Not found
WitnessAIWitnessAI is an AI security and governance platform that lets enterprises observe, control and protect employee and agent use of AI. (site, source)Not found
ZenityZenity is an AI agent security and governance platform that evaluates every agent action in real time. (site, source)Not found

AI security startups that raised funding recently

Rounds announced in the six months to October 19, 2026, newest first. Each links the company's record and its sources.

Which of these companies AI names

On October 19, 2026, Presspool asked the 8 questions below on ChatGPT, Claude, Perplexity, Gemini, Copilot, Google AI Mode and Google AI Overviews and recorded 53 answers. Not sampled: 3 Google AI Overviews answers. A company counts when an answer names it or cites its website. Answers vary from run to run, so each count is a dated sample, not a verdict.

CompanyNamed or cited inChatGPTClaudePerplexityGeminiCopilotGoogle AI ModeGoogle AI Overviews
Reco5 of 53 answers1 of 80 of 83 of 81 of 80 of 80 of 80 of 5
Harmonic Security2 of 53 answers0 of 80 of 81 of 81 of 80 of 80 of 80 of 5
Rein4 of 53 answers1 of 80 of 81 of 80 of 80 of 81 of 81 of 5
Opsin1 of 53 answers1 of 80 of 80 of 80 of 80 of 80 of 80 of 5
Outerlimit0 of 53 answers0 of 80 of 80 of 80 of 80 of 80 of 80 of 5
Kontext0 of 53 answers0 of 80 of 80 of 80 of 80 of 80 of 80 of 5
Palma.ai0 of 53 answers0 of 80 of 80 of 80 of 80 of 80 of 80 of 5
Giskard10 of 53 answers1 of 81 of 81 of 82 of 82 of 82 of 81 of 5
HiddenLayer27 of 53 answers3 of 82 of 85 of 86 of 85 of 83 of 83 of 5
Lasso Security11 of 53 answers2 of 81 of 83 of 83 of 80 of 82 of 80 of 5
Mindgard22 of 53 answers4 of 82 of 82 of 85 of 83 of 83 of 83 of 5
Noma Security13 of 53 answers2 of 82 of 84 of 81 of 80 of 82 of 82 of 5
Pillar Security8 of 53 answers2 of 82 of 81 of 82 of 81 of 80 of 80 of 5
Straiker10 of 53 answers2 of 82 of 82 of 81 of 80 of 81 of 82 of 5
WitnessAI8 of 53 answers2 of 81 of 82 of 81 of 80 of 80 of 82 of 5
Zenity15 of 53 answers2 of 83 of 84 of 83 of 82 of 81 of 80 of 5

Sources the AI apps cited most for these questions

  1. mindgard.ai: 10 of 53 answers
  2. microsoft.com: 8 of 53 answers
  3. openai.com: 8 of 53 answers
  4. paloaltonetworks.com: 8 of 53 answers
  5. linx.security: 7 of 53 answers
  6. straiker.ai: 7 of 53 answers
  7. checkpoint.com: 6 of 53 answers
  8. gartner.com: 6 of 53 answers
  9. respan.ai: 6 of 53 answers
  10. techcrunch.com: 6 of 53 answers
  11. tigera.io: 6 of 53 answers
  12. dealroom.co: 5 of 53 answers
  13. superblocks.com: 5 of 53 answers
  14. cbinsights.com: 4 of 53 answers
  15. crn.com: 4 of 53 answers

What counts as AI security

A company joins this map when its own site describes AI security as its main product and an independent source confirms it. Out of scope: broad security platforms where AI security is one feature, AI security units inside large security companies (including acquired startups), and companies whose AI defends other systems rather than securing AI itself. Definitions are versioned; this is version 1. Corrections need a public source: see Contact.

Questions this map tracks

Changes