Your AI-generated code ships in hours. An attacker’s agent can take it apart just as fast.
AI coding assistants have changed how software gets built. Code is generated, copied and shipped faster than teams can review it, often with no clear owner and little scrutiny of what it pulls in. Much of it runs in the browser, where anyone can inspect it.
Attackers have noticed. Adversarial AI agents can now analyze client-side code, extract the logic that matters and reverse engineer it at scale, and unreviewed AI-generated code is an easy target. The same agents are being used to replicate software outright, copying an application’s features, workflows and business logic without building any of it from scratch.
In this webinar, Pedro Fortuna, Jscrambler’s Co-Founder and CTO, Gareth Bowker, Head of Information Security, and Katia Kupidonova, Product Marketing Manager, break down this new attack surface using findings from Jscrambler’s latest research. They’ll explain where these threats are heading. They’ll also share insight into the new capabilities Jscrambler is building to make reverse engineering by LLMs costly.
What We Will Cover:
- The review gap: why ownership, review and dependency scrutiny aren’t keeping pace with AI-assisted development
- How adversarial agents work: how attackers use LLMs to analyze, extract and deobfuscate client-side code, and to replicate entire applications
- Where risk surfaces: why weaknesses introduced at build time only show up in live user sessions in the browser
- What comes next: where AI-driven attacks are heading, and insight into upcoming Jscrambler capabilities designed to make LLM-driven reverse engineering costly
