Telemetry data from the 2026 State of Browser Security Report reveals that, while the browser has become the de facto operating system for work in the enterprise, it remains one of the least secured segments in the overall security stack. In 2025, AI-native browsers, embedded copilots, and generative tools transitioned from being experimental pilots to being ubiquitous, routine tools for search, write, code, and workflow automation, thus creating a significant disconnect between the way employees are actually working and the organization’s risk monitoring capabilities.
The data also indicates that generative artificial intelligence has become an integral part of browser workflows, extending beyond the browser as a gateway for a small set of approved tools. According to the telemetry data collected by Keep Aware, 41% of end-users interacted with at least one AI tool on the web in 2025, with an average of 1.91 AI tools used per end-user, thus revealing the widespread integration of AI tools in the browser workflows. However, it has been observed that governance has not kept pace with the adoption of these tools, with end-users using their own accounts or unauthorized tools in the same browser session as their work activities.
This behavioral reality is especially dangerous when it comes to sensitive data exposure. In a one‑month snapshot of authenticated sessions, 54% of sensitive inputs to web apps went to corporate accounts, while a striking 46% went to personal or unverified work accounts, often within “trusted” apps like SharePoint, Google services, Slack, Box, and other collaboration tools. Because traditional DLP tools focus on email, network traffic, or endpoint files, they largely miss typed inputs, pasted content, and file uploads occurring directly inside live browser sessions, where today’s AI‑driven work actually happens.
Attackers have adapted to this shift as well, increasingly targeting the browser layer to bypass hardened email, network, and endpoint defenses. Keep Aware observed that 29% of browser‑based threats in 2025 were phishing, 19% involved suspicious or malicious extensions, and 17% were social engineering, highlighting how social and UI‑driven tactics dominate. Notably, phishing domains had a median age of more than 18 years, indicating adversaries are abusing long‑standing, seemingly trustworthy infrastructure rather than relying only on newly registered domains that filters are tuned to flag.
Browser extensions add another, often underestimated, attack surface. According to the report, 13% of unique installed extensions were rated High or Critical risk, meaning a significant slice of add‑ons running inside production environments have elevated permissions and potentially dangerous capabilities. Many extensions marketed as productivity tools request broad access to tabs, cookies, storage, and web requests, quietly gaining deep visibility into user sessions and sensitive business data without ongoing scrutiny.
The report makes a clear case that static controls—such as one‑time extension reviews, app allowlists, and domain‑based blocking—are no longer enough in a world of AI copilots, browser‑centric workflows, and adaptive phishing campaigns. Instead, organizations must treat the browse
[…]
Content was cut in order to protect the source.Please visit the source for the rest of the article.
[…]
Content was cut in order to protect the source.Please visit the source for the rest of the article.
This article has been indexed from CySecurity News – Latest Information Security and Hacking Incidents
Read the original article: