Mindgard is an AI security company focused on helping enterprises discover, assess, and defend AI systems, models, agents, and applications. Spun out of more than a decade of AI security research at Lancaster University and headquartered in Boston and London, it brings an attacker-aligned approach to securing generative AI and agentic environments across development, deployment, and runtime.
Mindgard treats AI security as a system problem rather than a model-only issue. Its platform maps the AI attack surface across models, prompts, tools, APIs, workflows, and connected data sources, then combines automated testing, runtime defenses, governance reporting, and expert services to surface exploitable weaknesses before they create business impact. The company is built for enterprises adopting AI at scale that need repeatable security validation, faster remediation, and stronger governance evidence.
Offerings, Capabilities, and Integrations
Mindgard’s capabilities span AI asset discovery, reconnaissance, adversarial testing, runtime detection and response, and governance reporting. Its attacker-aligned methodology is designed to reveal how real adversaries can probe, exploit, and chain failures across complete AI systems, including agents, guardrails, and downstream integrations.
The platform is designed to fit existing engineering and security operations. Mindgard supports web, CLI, Python SDK, and Burp Suite-based workflows, CI/CD pipeline checks, webhook-based export into ticketing and SIEM processes, and enterprise deployment options such as private tenant instances and SSO. Its documentation also shows preset support for common AI endpoint patterns including OpenAI, Azure, HuggingFace, and Anthropic.
Products and Services
- AI Discovery & Recon: Maps AI inventory and attack surface by identifying models, agents, tools, guardrails, connected infrastructure, and shadow AI so teams can target testing and risk analysis more effectively.
- AI Red Teaming: Continuously emulates real attacker behavior across models, agents, applications, data flows, and workflows to uncover high-impact vulnerabilities before they are exploited.
- AI Assessment: Validates AI security and safety posture using attacker-aligned datasets and multi-step testing to expose weaknesses in guardrails, prompts, access controls, and application behavior.
- AI Runtime Protection: Monitors live AI interactions, detects malicious activity in production, and applies context-driven protections and response measures to reduce active risk.
- Offensive Security: Mindgard’s flagship offensive testing offering, powered by an attacker-aligned attack library to pressure-test AI systems across development, deployment, and runtime.
- Model Scanning: Scans AI models for security vulnerabilities, policy violations, and harmful behaviors through offline profiling and run-time testing, with remediation guidance and risk analysis.
- AI Governance & Compliance: Centralizes AI risk analysis, inventory, and reporting while mapping findings to frameworks such as MITRE ATLAS, NIST, OWASP, and AIUC-1 to support governance and audit processes.
- AI Academy: Provides guided training, labs, and educational content to help teams build in-house AI security, adversarial testing, and secure AI deployment skills.
- Mindgard Services: Umbrella advisory and enablement services that give customers direct access to AI security specialists for assessment, testing acceleration, and capability building.
- AI Red Teaming & Pentesting as a Service: Expert-led offensive testing delivered with specialist support to complement the platform’s automated AI security capabilities.
- Technical Account Management: Dedicated operational support service that helps customers deploy, automate, and get ongoing value from the Mindgard platform.
- AI Security Expert: Subscription service that gives customers access to an offensive security specialist for posture assessments, tooling improvement, and adversarial skill development.
- AI Security Training: Expert-led instruction covering AI security fundamentals, adversarial techniques, and secure AI practices for practitioners and security teams.
Target Customers
Mindgard targets enterprises that build, buy, or operate AI models, agents, and AI-enabled applications in production. Its core users span security, application security, red team, AI/ML engineering, platform engineering, and governance, risk, and compliance functions that need evidence-based validation of AI exposure rather than model-only testing.
Customer evidence points to fit in large and complex organizations, including insurance, healthcare, biopharmaceutical, semiconductor, financial, and enterprise technology environments. The platform is well suited to organizations running internal copilots, customer-facing AI services, or sensitive AI workflows where continuous testing, runtime monitoring, and audit-ready reporting are operational requirements.
Cloud Integrations and Marketplace
- Microsoft Azure: Mindgard documentation includes preset templates and configuration examples for Azure-hosted AI endpoints, including Azure deployments for GPT-style models, indicating support for testing Microsoft Azure-based AI environments.
Key People
- James Brear: Chief Executive Officer
- Dr. Peter Garraghan: Chief Science Officer and Founder
- Aaron Portnoy: Chief Product Officer
- Sasha Polishchuk: VP Engineering
- Fergal Glynn: Chief Marketing Officer
- Rich Smith: Head of Research
Key Facts
- Headquarters: Boston, Massachusetts, USA
- Employees: Approximately 29
- Annual Revenue: Undisclosed
- Parent Company: None
- Subsidiaries: None
- Publicly Listed: No (privately held)
Analyst Recognitions
- Gartner: 2026 Emerging Tech: Top-Funded Startups in AI TRiSM: Agentic AI and Beyond – included among startups advancing AI security testing. 2025 Gartner Hype Cycle for Application Security – Representative Vendor in AI Security Testing.
