AI Agent & LLM Security
7 providers in this category.
Agent and LLM security platforms defend AI applications against prompt injection and data leakage, and test agents and models for exploitable weaknesses before attackers do. Capabilities split across red teaming, runtime protection (inline blocking of live traffic), and model supply-chain scanning. The category consolidated sharply in 2025-2026 via acquisition; pricing is almost universally enterprise quote-based.
Companies
Neutral ordering. Not a recommendation.
Enkrypt AI
www.enkryptai.com
Enkrypt AI is a full-lifecycle AI security, compliance, and governance platform for LLMs, AI agents, multimodal systems, and MCP workflows, covering red teaming, risk discovery, and runtime guardrails against prompt injection, data leakage, and adversarial attacks.
Giskard
www.giskard.ai
Giskard is an open-source testing and red-teaming framework for evaluating machine learning models, LLMs, and RAG pipelines against vulnerabilities like prompt injection and hallucination, founded in 2021 and headquartered in Paris, France.
HiddenLayer
hiddenlayer.com
HiddenLayer is an enterprise AI security platform focused on ML model supply-chain scanning, adversarial red teaming, and runtime prompt-injection defense, founded in 2022 and headquartered in Austin, Texas.
Mindgard
mindgard.ai
Mindgard is a continuous, automated AI red teaming platform for testing LLMs and AI agents against adversarial attacks, aligned to the MITRE ATLAS framework, spun out of AI security research at Lancaster University and formally founded in 2022 in London.
Noma Security
noma.security
Noma Security is an AI security platform covering discovery, security posture management, automated red teaming, and runtime protection for enterprise AI applications, agents, RAG pipelines, and MCP servers, founded in 2023 and headquartered in Mountain View, California.
Pillar Security
www.pillar.security
Pillar Security is a unified AI security platform providing visibility, risk assessment, and protection across the AI lifecycle, discovering AI assets including agents, models, prompts, and tools and mapping attack surfaces via its RedGraph capability for agentic red teaming.
WitnessAI
witness.ai
WitnessAI is an inline "AI firewall" that provides agentless network-level discovery of AI usage plus an intent-based policy engine for inline protection, including PII tokenization and prompt-injection blocking, founded in 2023 and headquartered in Mountain View, California.