In this interview with Help Net Security, Rui Ribeiro, CEO of Jscrambler, explains why the browser has become a security problem organizations do not control. Companies do not own the device, the extensions, or the network path, yet that is where application logic, third-party code, customer data, and AI meet during every customer interaction.
He discusses the limits of Content Security Policy and Subresource Integrity, the risks of third-party AI chat scripts running with the same privileges as the application, and what regulators expect when they ask what executed inside a user’s session. He also argues that AI lowers the cost, time, and expertise attackers need.
Server-side security rests on controlling the runtime. In the browser you control none of it: not the device, not the extensions, not the network path. How should defenders reason about a runtime that belongs to someone else?
<a href="https://bitcomme.com/traditional-defense-companies-invest-frequently-in-tech-startups/” title=”Traditional defense companies invest frequently in tech startups”>Traditional security assumes organizations can secure the environments they own. The browser challenges that assumption. It’s where proprietary application logic, third-party software, customer data, and increasingly AI all come together during every customer interaction. Organizations can no longer rely solely on trusting what loads into the browser. They must also govern what software executes, what data it can access, what actions it can perform, and whether those actions align with business policy. Browser security is no longer just about protecting web applications against client-side threats. It’s about ensuring that software, data, and AI behave as intended throughout every customer interaction.
Content Security Policy is more than a decade old. So is Subresource Integrity. Why have those controls not closed the gap?
CSP and SRI remain important security controls, but they solve a different problem. They help establish trust before software executes by validating where code comes from and whether it has been modified. Today’s risk increasingly emerges after execution begins. Once trusted scripts are running, organizations still need visibility into what data they access, what actions they perform, how AI changes their behavior, and whether they continue operating within least-privilege boundaries. Traditional controls establish trust. Modern browser security governs execution.
Agentic browsing removes the assumption that a human is behind the mouse. When an autonomous agent walks through your checkout, what breaks first: fraud controls, consent capture, or session logic?
Agentic AI changes much more than checkout flows. It fundamentally changes the economics of attack. Capabilities that once required highly specialized reverse-engineering expertise can increasingly be accomplished by capable models responding to well-crafted prompts. We’ve already seen attackers use AI to rapidly understand application logic, replicate digital experiences, and accelerate fraud. The challenge is recognizing that AI lowers the cost, time, and expertise required to compromise browser-based applications.
Companies are dropping LLM chat widgets into production pages, often delivered by a third-party script. What is the worst, realistic outcome of that pattern?
The greatest risk is the execution model. Third-party AI scripts often operate with the same browser privileges as the rest of the application, giving them visibility into far more than the conversation itself. Without proper runtime governance, they may access payment information, account details, proprietary workflows, pricing logic, or sensitive customer data elsewhere on the page; all before any sort of data submission.
Once that information leaves the browser and enters external AI services, organizations lose practical control over how it’s processed, retained, or reused, creating simultaneous privacy, compliance, intellectual property, and competitive risks.
Regulators and litigators increasingly ask organizations to demonstrate what executed inside a user’s session. Most cannot answer. What does a usable evidence trail look like when a company does have one?
A meaningful evidence trail goes well beyond maintaining an inventory of scripts. It should demonstrate what software executed, what data was accessed, what policies were enforced, what behaviors were blocked, and whether third-party code operated within defined privileges throughout the session. Increasingly, organizations won’t just need visibility into browser activity; they’ll need evidence that governance was actively enforced while customer interactions occurred.
Some readers will hear “the browser is the new edge” as a vendor inventing a category. Make the strongest case against your own position and tell us what evidence would change your mind.
The strongest counterargument is that browser security isn’t new. Organizations have long relied on secure development practices, CSP, SRI, application security testing, and endpoint protections to reduce client-side risk. If those approaches fully addressed today’s challenges, however, we wouldn’t continue seeing third-party compromises, client-side fraud, privacy violations, AI data leakage, and compliance failures originating inside live browser sessions.
The browser hasn’t suddenly become important; it has fundamentally changed. It’s now where software is assembled, where sensitive data is created, and where AI increasingly interacts with users. Browser security is an existing discipline whose importance has expanded.
What assumption in current browser security thinking will look wrong in three years?
The biggest assumption that will prove wrong is that browser security is solely an application security initiative. Increasingly, software supply chain security, data privacy, AI governance, fraud prevention, compliance, and digital trust all converge in the browser because that’s where software executes, data is created, and AI operates.
Organizations that continue treating browser security as a narrow AppSec problem will struggle to address risks that now span multiple executive stakeholders. Browser security will increasingly become a shared business initiative, not because the browser changed, but because everything businesses care about now intersects there.
Learn more: Jscrambler launches Unified Client-Side Security Platform
