Why AI-era attacks demand deterministic defense

In 2026, the security industry is grappling with the rapid acceleration of vulnerability discovery driven by AI tools like Anthropic’s Mythos and OpenAI’s Daybreak, which has led to a sharp increase in reported vulnerabilities and is overwhelming security teams. Despite the allure of using AI to counter AI-driven attacks, the real issue lies in the gap between identifying vulnerabilities and acting on them, largely due to the constraints of legacy systems and existing operational workflows. The key to effective defense is not merely speeding up discovery but improving the design of security systems to integrate deterministic workflows that can execute automatically and consistently. This approach, as exemplified by the well-defended curl codebase, emphasizes building resilient systems that inherently reduce exploitable surfaces, rather than relying on real-time improvisation. By embedding AI in the design phase to develop smarter defenses and utilizing human oversight for critical decisions, organizations can better manage threats and maintain security without being dependent on continuous, costly AI reasoning.