Ex-Google Security Execs Just Raised $36 Million to Fight AI-Generated Spear Phishing
Here’s something that keeps me up at night: AI-generated phishing emails that know your name, your projects, your travel plans, and exactly which tone of voice your boss uses. They don’t have typos. They don’t have the telltale formatting issues we’ve all learned to spot. They’re personalized, contextual, and they work.
This isn’t hypothetical anymore. It’s happening right now, and traditional email security isn’t catching it.
That’s the gap that AegisAI is trying to fill. The startup, founded by former Google security executives Cy Khormaee and Ryan Luo — the same team that built safe browsing technology and reCAPTCHA — just announced a $36 million Series A round led by Battery Ventures, with participation from Accel and Foundation Capital. The round brings AegisAI’s total funding to $49 million less than a year after its launch.
The Problem: Email Security Built for Yesterday’s Threats
The co-founders spent a decade preventing email hacks at Google. What they saw convinced them that the old approach just doesn’t work anymore.
Conventional email security relies on rule-based systems — “if-then” logic that checks for known patterns, suspicious domains, and blacklisted senders. The problem is that AI-generated phishing emails don’t fit those patterns. They’re unique every time. They don’t repeat the same payload signatures or use the same subject lines. Traditional filters might as well be trying to catch smoke with a net.
The numbers back this up. Khormaee told TechCrunch that AI-powered attacks now bypass existing controls more than half the time — making them almost twice as effective as they used to be. “They’ve researched you, they understand everything about you, and they’re targeting attacks that are perfectly bespoke to you,” he said.
Consider what an AI can do with publicly available information. It can scrape your LinkedIn profile, your company’s blog, your recent conference presentations, and your team’s Slack status messages. Then it can craft an email from your CTO asking you to approve an urgent invoice, using the same terminology, the same writing style, and referencing a project you’re actually working on. That’s not a generic phishing template anymore — that’s a targeted social engineering attack delivered at machine scale.
How AegisAI’s Approach Is Different
Instead of relying on rules, AegisAI built AI agents that analyze each incoming message the way a human security analyst would. The models pay attention to subtle anomalies — small inconsistencies in language, unusual request patterns, deviations from normal communication behavior — that even the most thorough checklist wouldn’t catch.
For example, the startup’s AI can catch malicious PDF attachments that look legitimate at first glance, including ones with built-in passwords and CAPTCHA that are designed to fool standard spam filters. It’s the kind of attack that would sail past a traditional gateway because the PDF itself is technically clean — it’s the context around it that’s suspicious.
This agentic approach is what sets AegisAI apart from legacy vendors like Proofpoint and Mimecast, which still rely heavily on signature-based detection. It also puts the startup in direct competition with newer players like Ocean (backed by Lightspeed, with $28 million in funding) and Abnormal Security. But Dharmesh Thakker, general partner at Battery Ventures, believes AegisAI has an edge because it’s led by the people who helped secure Gmail — the most widely used email system in the world.
Early Traction and What Comes Next
AegisAI already has dozens of customers, including crypto payments company Mesh, AI startup LangChain, and privacy compliance platform Lokker. That’s solid traction for a company that’s less than a year old and competing in a space where enterprises are famously slow to switch vendors.
The startup’s ambition doesn’t stop at email. Khormaee indicated that the company plans to expand into other areas of defense, including data security. “The core idea of building customized, highly advanced agents that can do investigations is going to determine who becomes the next dominant security company,” he told TechCrunch.
That broader vision makes sense. Once you’ve built an AI agent that can analyze email traffic for anomalies, the same architecture can be applied to cloud logs, endpoint telemetry, or network traffic. The question is whether AegisAI can execute on that vision before the established players catch up with their own AI-native approaches.
Why This Matters for Every Developer and Tech Professional
I’ve written before about how AI attacks evolve faster than our defense-in-depth strategies can adapt. I recently covered how a single malicious email can plant false memories in an AI assistant through the MemGhost attack, which weaponizes the very channel we trust most — email — against next-generation AI tools. AegisAI is essentially trying to solve the same problem from the defense side: building AI systems that can recognize when they’re being manipulated, whether the target is a human or another AI.
There’s also a growing trend of phantom squatting, where AI hallucinates domain names that attackers can register and use for phishing campaigns. This isn’t just about email anymore — it’s about AI-generated attack surfaces that didn’t exist two years ago.
On the defense side, projects like Cisco Antares are pushing compact, open-source AI models designed specifically for security tasks, making it cheaper and easier for organizations to deploy AI-powered defenses. AegisAI’s approach fits into this larger shift: security is becoming an AI-vs-AI battlefield, and the winners will be the ones who build the most effective agents.
Meanwhile, the scale of AI-driven attacks continues to grow. The FakeGit campaign showed that attackers are willing to create thousands of malicious repositories leveraging AI-generated code to slip malware past unsuspecting developers. The pattern is the same: AI lowers the cost of creating targeted, believable attacks, and the defenders need AI just to keep pace.
The Bottom Line
Email has been the most exploited attack vector for as long as I’ve been working in tech. What’s changing is the sophistication. AI hasn’t just made phishing more efficient — it’s changed the nature of the threat. A targeted spear phishing email that used to take a skilled attacker hours to research and write can now be generated in seconds.
Startups like AegisAI represent a necessary evolution in how we think about security. The old model of blocking known bad things doesn’t work when every attack is custom-made. The new model has to be about understanding intent, context, and behavior — which is exactly what AI agents are good at.
Will AegisAI become the next dominant security company? That depends on execution, timing, and whether it can expand beyond email before the market shifts. But one thing is clear: the AI-vs-AI arms race in cybersecurity is here, and it’s only going to accelerate.