ReadingResearchRadarInvestment framework
Sign in / Sign up
← Back to reading
News3 min read·2026-10-19

SailPoint Founder Mark McClain: AI Security Is Like F1 Brakes — Better Stopping Means Going Faster

SailPoint founder Mark McClain argued during a recent visit to Hong Kong that AI security incidents should not become a reason to slow development, but rather a catalyst for building real-time security architectures. He used an F1 racing analogy to explain that good brakes are not an obstacle to speed but a prerequisite for drivers to go all out. He introduced the concept of 'adaptive identity,' arguing that enterprise identity management must shift from processing changes on an hourly or daily basis to microsecond-level real-time responses to handle the dynamic environment created by AI agents.

ImportanceMaterialEvidenceE3 inspectableWrite-upQuick

2026-10-19 00:12 GMT+8

Recent AI model service failures have sparked industry debate: should AI development slow down? Mark McClain, founder and CEO of identity security company SailPoint, gave a clear negative answer during his Hong Kong visit. His core argument: slowing AI adoption only makes it easier for malicious actors to succeed. The real path forward is not deceleration but building security defenses that can keep pace with AI agents.

McClain used an F1 racing analogy to explain this logic. A race driver sprints into a corner at 200 miles per hour only if he trusts his braking system. If the brakes are unreliable, the driver won't go full speed into the turn and cannot win the race. The business world is the same: enterprises need a powerful engine (AI capability) to go fast, but when they hit a corner (a security incident), they need excellent brakes (security protections) even more to keep moving forward.

Behind this analogy lies a specific technical judgment. McClain pointed out that existing identity management mechanisms were built in an era that did not account for AI agents. Under the traditional model, new identity provisioning and existing identity changes are processed on an hourly, daily, or weekly basis. But agentic AI has changed this rhythm — when an AI agent or authentication faces a challenge or is compromised, changes and restrictions must happen at the microsecond level. He described this shift as a paradigm move from 'administratively oriented' to 'real-time dynamic management.'

Sign in / Sign up
ReadingResearchRadarInvestment framework

His proposed solution is the 'adaptive identity' concept that SailPoint is pushing: based on context and intent, it assesses in real time whether a specific identity is exceeding its boundaries and takes immediate blocking measures. McClain acknowledged that SailPoint's past products were management-oriented, making changes only at human-processing speed; the company is now driving a paradigm shift at both the operational and product levels, with multiple products deploying AI agents to detect anomalies and adjust in real time.

McClain also acknowledged an inherent tension in this approach. Controlling AI agents is difficult and will relatively limit their capability, because any innovative technology's development and adoption must balance user convenience against security. He believes the key is providing 'frictionless' security — not fully frictionless, but as frictionless as possible — so users can continue using AI while the system detects boundary crossings and intervenes.

From an enterprise perspective, McClain observed that all companies are caught in this contradiction: on one hand they feel pressure to accelerate AI adoption, on the other, incidents like the Hugging Face breach and data leaks keep erupting, highlighting that certain controls are not yet in place. He predicts that as enterprises expand their deployment of enterprise-grade AI — for example, using AI to optimize business processes — those deployments will face extremely strict review. Companies are, in his words, at 'the edge of the edge.'

It is worth noting that McClain's role as founder of an identity security company means his argument channels the security need directly into his own product category. 'Adaptive identity' currently exists only as a concept description and early proof-of-concept narrative, with no publicly available independent performance data or deployment-scale verification. His F1 analogy precisely reframes the relationship between security and speed, but whether that relationship holds in actual enterprise environments still depends on whether protection mechanisms can achieve microsecond-level response without significantly reducing AI agent efficiency.

Sources

  • 原始出处ithome.com.tw↗
Continue researching →