Glow raises $1.2B to rethink endpoint security for AI agents
The race to secure AI-powered workplaces has a new heavyweight. Glow Inc. is emerging from stealth with a $1.2 billion valuation and a pitch that traditional endpoint security tools can’t keep up with the risks introduced by AI agents and developer toolchains.
Founded by veterans who previously built endpoint security platforms at companies like CrowdStrike and SentinelOne, Glow argues that current antivirus and EDR solutions were designed for yesterday’s threats—static binaries and human-driven attacks—not the dynamic, code-executing agents now roaming enterprise networks. Its platform focuses on monitoring and governing AI models and the tools that deploy them, aiming to prevent malicious use before damage occurs rather than after.
A new threat surface
Glow’s approach highlights a gap many CISOs have only recently begun to acknowledge: AI agents don’t behave like traditional software. They can spawn sub-agents, chain API calls across services, and modify their own code—behaviors that bypass signature-based defenses and anomaly detection tuned for human activity. The startup claims its runtime monitoring can trace agent behavior in real time, flagging policy violations such as unauthorized data exfiltration or unapproved tool usage.
The investor bet
The $1.2 billion valuation comes on the heels of a $100 million Series B led by a16z, with participation from Index Ventures and existing backers like Accel. Investors are betting that securing AI-native infrastructure will become a multi-billion-dollar market as enterprises embed hundreds of autonomous agents in their workflows. Glow’s backers also point to the company’s technical depth—its founders built detection engines used at scale in prior endpoint companies—as a reason for the aggressive valuation.
Why it matters
Glow’s arrival signals that endpoint security is entering a new phase, where the adversary isn’t just malware but also rogue AI behavior. For security teams, the shift demands tools that understand code execution paths and policy intent, not just file hashes and registry changes. If Glow succeeds, it could redefine what “endpoint” means in an AI-first enterprise—placing governance at the heart of security rather than treating it as an afterthought. The real test will be whether its runtime approach can scale across the messy reality of today’s AI deployments without bogging down developers.
Source: TechCrunch. AI-assisted editorial synthesis — TechnoExpress.