A Security Raises $37M to Counter AI-Driven Cyberattacks

A Security Emerges to Fight Machine-Speed Threats

A New York-based cybersecurity startup has raised $37 million to address what its founder calls a "generational risk": attacks conducted entirely by AI systems operating at machine speed and scale.

A Security, which emerged from stealth this week after launching in January 2025, secured funding from Lightspeed Ventures Partners and Cyberstarts to build defenses against adversaries that can automate discovery, exploitation, and attack-path analysis far beyond human capabilities. The company was founded and is led by Yossi Torati, who previously spent nearly six years at Israeli cyber consulting firm Sygnia, where he led enterprise security operations.

Why it matters

Organizations are deploying AI agents and applications at unprecedented speed due to business pressure to innovate, but security teams lack tools designed for threats that operate continuously across multiple attack surfaces. Traditional patch management cannot keep pace with AI-driven attacks that identify and exploit vulnerabilities within minutes. This funding signals investor recognition that autonomous defensive systems will become necessary as attackers weaponize frontier AI models.

The Threat Landscape Shifts to Machine Adversaries

Torati told Government Info Security that frontier AI models are rapidly improving at cybersecurity tasks, creating an urgent need for organizations to defend against these capabilities in adversarial hands. Unlike human attackers who work sequentially and require rest, AI-powered threats can operate continuously while evaluating massive amounts of information simultaneously.

The threat extends beyond traditional code vulnerabilities. Because AI interfaces rely heavily on natural language, attackers can manipulate an AI system's reasoning process through prompts and interactions—essentially conducting social engineering against machines rather than people. Torati noted that attackers can manipulate AI agents into violating policies, exposing information, or performing unauthorized actions.

Unified View Across Attack Surfaces

A Security's platform evaluates relationships between environments to identify realistic attack paths that could lead to business impact. The system analyzes traditional vulnerabilities, zero-day exploits, and misconfigurations across web applications, AI applications, APIs, and networks.

Torati emphasized that organizations need a unified view because modern attackers operate across multiple attack surfaces rather than targeting systems in isolation. AI systems can retain and analyze far more context than humans, making them particularly effective at identifying business logic flaws and sophisticated attack chains that might otherwise remain hidden.

Autonomous Remediation as the Answer

The company's roadmap focuses on building what Torati calls "the autonomous remediation company." He expects the industry will move toward systems that automatically make configuration changes, apply compensating controls, and reduce risk without waiting for manual approval.

While organizations may initially hesitate to grant this level of autonomy to security systems, Torati believes businesses will eventually face a choice between allowing AI-driven attacks to succeed or permitting automated defensive actions. When weaponized AI becomes prevalent, he argued, security teams cannot rely solely on patching—they must develop the ability to rapidly apply compensating controls while remediation efforts are underway.

"You cannot outpace weaponized AI by patching," Torati said, noting that future attackers will operate too quickly for traditional patch management to provide sufficient protection.

These details were first reported by Government Info Security.