Adversa AI

Adversa AI is a specialist AI red teaming company that helps organizations identify and remediate vulnerabilities in AI agents, large language models, and generative AI applications through simulated adversarial attacks. Founded in 2021 in Tel Aviv by Alex Polyakov, the company built early credibility through original research on adversarial machine learning, LLM jailbreaks, and AI trust, publishing widely cited work on securing AI systems. Adversa's offering combines automated and expert-led red teaming with AI security awareness, assessment, and assurance services, covering the full spectrum from model evasion and prompt injection to agent-level exploitation. Its continuous AI red teaming platform is designed to test agentic and generative systems on an ongoing basis rather than through a single penetration test, mapping findings to attacker techniques and providing remediation guidance. Adversa targets enterprises, AI builders, and security teams that need deep, research-driven adversarial testing of AI systems. As a research-forward boutique, its distinguishing characteristic is domain expertise in adversarial AI and a threat-intelligence orientation, making it a fit for organizations that want specialist attack simulation rather than a broad general-purpose security suite.