AI Red Teaming & Stress-Testing

Adversa AI

Tel Aviv, Israel

Adversa AI focuses on securing artificial intelligence systems against cyber threats, privacy risks, and safety incidents. Based in Tel Aviv, the company provides AI threat modeling, security architecture review, adversarial testing, and awareness training for organizations deploying machine learning models across regulated industries including finance, healthcare, and critical infrastructure.

Bluefire Redteam

New York, United States

Bluefire Redteam is a cybersecurity firm specializing in AI and LLM application penetration testing alongside traditional red team services. The company simulates real-world threats including prompt injections, jailbreaks, and logic bypasses against applications powered by large language models. Additional services include purple team testing, source code review, and application security consulting.

CrowdStrike AI Red Team Services

Austin, United States

CrowdStrike offers dedicated AI red team services providing in-depth evaluations of large language model applications tested against the OWASP Top Ten. Based in Austin, Texas, the cybersecurity leader simulates real-world attacks tailored to each organization's specific AI use cases, combining offensive red team exercises with defensive blue team detection and response capabilities.

HackerOne AI Red Teaming

San Francisco, United States

HackerOne provides AI red teaming services combining expert human researchers with agentic testing tools to evaluate AI systems for jailbreaks, prompt injection, misalignment, and policy violations. Based in San Francisco, the platform delivers validated findings mapped to OWASP LLM Top 10, MITRE ATLAS, and NIST AI RMF frameworks with prioritized remediation recommendations.

HiddenLayer

Austin, United States

HiddenLayer is an AI security company based in Austin, Texas, providing automated red teaming, runtime protection, and policy enforcement for enterprise AI deployments. The platform detects prompt injection, model extraction, and unauthorized tool usage by operating independently of the AI model's reasoning process, ensuring deterministic security controls hold even if the model is compromised.

Lakera

Zurich, Switzerland

Lakera is an AI-native security platform based in Zurich, Switzerland, trusted by Fortune 500 companies to accelerate generative AI initiatives. The platform prevents prompt injections, data leakage, and jailbreaks in real time with context-aware detection that adapts to evolving threats. Lakera also created Gandalf, the most popular cybersecurity game for AI security education.

Mindgard

London, United Kingdom

Mindgard is an AI security company headquartered in London and Boston, specializing in automated red teaming and adversarial testing for machine learning models, LLMs, and AI agents. Founded at Lancaster University, the platform enables enterprises to identify, assess, and mitigate AI-specific vulnerabilities including prompt injection, data leakage, and model manipulation across production environments.

Protect AI

Seattle, United States

Protect AI operates Recon, a scalable red teaming platform for AI applications that systematically tests across multiple threat vectors using an attack library of over 450 known AI exploits. Based in Seattle, the company also runs huntr, the first bug bounty platform for AI and ML, backed by a community of over 17,000 security researchers.