How LangChain Stopped Thousands of AI Phishing Attacks
LangChain is the leading open-source framework for building applications powered by large language models. With a large developer community and high-profile engineers and executives operating in the public eye, LangChain is one of the most visible companies in AI making it a prime target for sophisticated, AI-generated email attacks.
Overview
Challenges
Hundreds of engineers publicly identifiable | Surge in AI-generated phishing emails | Fake invoices, gift card requests | No scalable email security in place
Solutions
AegisAI's AI-native email security platform | Deployed in under 5 minutes with zero disruption and zero rules or tuning needed | Real-time threat monitoring and per-user attack visibility | Always-on protection that scales with the team
The Challenge: Fighting AI With AI
When Arthur Stromquist joined LangChain as Security and Compliance Manager, one reality was immediately clear: the company's public profile was a liability. Engineers, researchers, and founders regularly publish papers, speak at conferences, and engage on social media making their email addresses easy to find and exploit.
"A lot of our engineers are more publicly facing," Arthur explains. "Just given the nature of our company, they get a huge rate of phishing attempts sent to them."
What was getting through
The threats weren't garden-variety spam. Attackers were using AI to craft convincing, personalized messages, fake emails impersonating CEO Harrison Chase requesting gift cards, fraudulent contract attachments, and targeted spear-phishing aimed at engineers with production system access. The volume and sophistication were accelerating, and LangChain had no scalable, low-friction email security in place.
"We were getting a lot of phishing attempts. We get our classic Harrison Chase emails—our CEO asking for a gift card. We see those non-stop. We've seen fake contracts come in and quite a few things just like that."
The Solution: Zero Friction, Maximum Coverage
Arthur reached out to Ryan, LangChain’s Head of IT, to evaluate options. After reviewing the landscape, including legacy vendors and newer entrants, they landed on AegisAI.
"Email protection is one of those gaps that a lot of companies can have," Ryan explains. "And it's the easiest attack vector to seal up. How you do it depends on what you choose and AegisAI presented us with an easy, low-friction implementation."
Deployment
Ryan set up AegisAI in under five minutes. No professional services engagement, no lengthy configuration, no disruption to the team. Critically, they didn't write a single rule or tweak a single setting. It worked out of the box.
“Ryan set it up within five minutes. That’s it. It just works,” Arthur says. “If you just need something that’s easy to deploy, low touch, and that just works, AegisAI is great because it scales.”
The Results: Thousands of Threats Caught Automatically
Within the first week of deployment, AegisAI began surfacing threats that the team had never had visibility into before. The breadth and sophistication of what was being flagged—and the intelligence provided about each attempt was eye-opening.
“I was astounded by all of the emails it was able to flag,” Ryan recalls. “The insights it gave us into the different threats that were coming after us, seeing exactly who’s being targeted, how often, and from where was incredible.”
Visibility
For Arthur, the operational value goes even deeper. AegisAI doesn’t just block threats it provides the visibility needed to proactively educate and protect the most at-risk employees.
Within the first week, AegisAI began surfacing threats the team had never had visibility into. The breadth of what was flagged and the intelligence provided about each attempt changed how the security team operated.
“With AegisAI, I can track who’s getting attacked and how much they’re getting attacked, and arm them to be protective of everything on their side,” he says. “We’ve found thousands of attempts.”
What AegisAI detected
- Executive impersonation emails (CEO, founders) requesting wire transfers and gift cards
- AI-crafted spear-phishing targeting engineers with production access
- Fraudulent contract attachments designed to compromise credentials
- High-volume opportunistic phishing exploiting publicly listed employee emails
Per-user analytics
For Arthur, the value goes beyond blocking threats. AegisAI provides per-user attack analytics, giving the security team visibility into who is being targeted most—and the ability to proactively educate those employees.
"With AegisAI, we can track who's getting attacked and arm them to be more protective of everything on their side," Arthur says. "We found thousands of attempts."
The Business Impact: Security That Disappears Into the Background
Perhaps the most telling measure of success is how little the LangChain team thinks about email security now. For a lean, fast-moving company, removing an entire category of operational overhead is significant.
“We very rarely have to work on email security from this angle anymore,” Arthur says. “We have AegisAI protecting us, so we can go focus on other tasks.”
Less to think about
Ryan echoes this sentiment when evaluating AegisAI against the broader vendor landscape, including legacy providers and newer competitors operating in the same space.
Why LangChain Chose AegisAI
For an AI company that lives at the frontier of what’s possible with Large Language Models, choosing a security partner required more than checkbox compliance. It required a vendor that could match both the sophistication of the threat and the pace of the company.
AI-native defense: Only AegisAI could fight AI-generated attacks with equivalent intelligence
Zero configuration: Worked out of the box—no rules to write, no settings to tweak, no tuning required
Speed of deployment: Under 5 minutes from sign-up to active protection—no professional services required
Operational simplicity: Zero ongoing maintenance burden for a lean security team
Deep threat visibility: Per-user attack analytics that inform proactive education
Founder-led partnership: Direct access to product and engineering leadership for rapid iterationProven scalability: Protection that grows with headcount and public surface area
See what AI-native security feels like
Thirty minutes. Real attacks pulled from environments like yours (BEC, vendor fraud, credential phishing) with the reasoning behind each verdict. No setup required.
Five-minute connect · monitoring mode · disconnect any time without touching mail routing