Mindgard

Active, privately held.

Sources 1, 2, 3 · checked 2026-11-08

• Automated AI red teaming with autonomous attack agents that run reconnaissance and multi-turn attacks; reruns tests in CI/CD on every change
• Expert-led AI red teaming services by human researchers with a named lead from scoping through retest
• Findings mapped to OWASP Top 10 for LLM Applications and MITRE ATLAS technique IDs
• Agent testing: tool poisoning, indirect prompt injection, privilege escalation across tools, MCP server abuse, data extraction
• Deliverables: attack narratives, reproducible transcripts, severity-ranked findings, control-effectiveness verdicts, remediation roadmap; successful attacks kept as regression tests
• Runtime threat detection and response

Sources 1, 2, 3 · checked 2026-11-08

• Teams that want an expert engagement followed by continuous automated regression testing
• Teams that need OWASP and MITRE ATLAS IDs in findings for auditors

Sources 1, 2, 3 · checked 2026-11-08

Contact vendor (book a demo)

Sources 1, 2, 3 · checked 2026-11-08

• SOC 2 Type 2 (badge on homepage)

Listed as stated by the vendor. Scope can differ by product or subsidiary; request current reports.

Sources 1, 2, 3 · checked 2026-11-08

• No public pricing
• Vendor states a single AI application engagement typically takes four to six weeks

Sources 1, 2, 3 · checked 2026-11-08