Cyberhaven research reveals rapid enterprise adoption of ChatGPT Atlas while highlighting critical browser agent security risk. Prompt-injection vulnerabilities threaten corporate data, emphasizing the need for caution and proactive security measures.
-- Cyberhaven, a leader in data security intelligence, has released new research uncovering the unprecedented adoption of ChatGPT Atlas, OpenAI’s newly launched agentic browser, alongside critical browser agent security risk findings that pose significant threats to corporate data integrity.
The research highlights that while ChatGPT Atlas is rapidly becoming one of the most widely adopted AI-powered browsers across enterprises, it introduces a browser agent security risk that organizations may not yet be prepared to manage effectively.
Unprecedented Adoption Across Industries
Cyberhaven’s analysis shows that 27.7% of enterprises already have at least one employee using ChatGPT Atlas, with some organizations reporting usage by as much as 10% of their workforce. The browser has been installed on 1.7% of corporate macOS endpoints, or approximately 17 out of every 1,000 devices.
The technology sector leads in adoption, representing 67% of users, followed by pharmaceuticals (50%) and finance (40%). Notably, ChatGPT Atlas has achieved 62 times more downloads in corporate environments compared to Perplexity’s Comet browser, despite Comet’s earlier launch in July 2025. Interestingly, Atlas’s debut also triggered a sixfold increase in Comet downloads, signaling accelerating interest in agentic browsers across the enterprise landscape.
Critical Browser Agent Security Risk Identified
While adoption has surged, Cyberhaven Labs warns that this enthusiasm comes with significant browser agent security risk concerns. During controlled red-team exercises, the company identified prompt-injection vulnerabilities within ChatGPT Atlas that could allow attackers to manipulate the browser’s behavior or exfiltrate sensitive data from corporate environments.
OpenAI’s Chief Information Security Officer, Dane Stuckey, has previously stated that “prompt injection remains a frontier, unsolved security problem.” Cyberhaven’s findings reinforce that agentic browsers, while powerful, represent a browser agent security risk because they operate autonomously — making decisions, executing tasks, and interacting with enterprise data in ways traditional browsers never have.
A Critical Moment for Enterprise AI Adoption
The timing of this research is crucial as enterprises accelerate their use of AI-powered tools to improve productivity. However, agentic browsers like ChatGPT Atlas introduce new layers of automation and autonomy that blur the line between user action and AI execution — a combination that expands the browser agent security risk surface dramatically.
About Cyberhaven
Cyberhaven is a data security company pioneering Data Behavior Analytics (DaBA) to protect enterprise information from insider threats, data loss, and emerging AI-related vulnerabilities. By analyzing how data moves, transforms, and interacts within organizations, Cyberhaven helps enterprises maintain visibility and control in an increasingly AI-driven world.
Contact Info:
Name: Samantha Rotbart
Email: Send Email
Organization: Cyberhaven
Website: https://www.cyberhaven.com/blog/browser-agent-security-risk-chatgpt-atlas
Release ID: 89175558
In the event of encountering any errors, concerns, or inconsistencies within the content shared in this press release, we kindly request that you immediately contact us at error@releasecontact.com (it is important to note that this email is the authorized channel for such matters, sending multiple emails to multiple addresses does not necessarily help expedite your request). Our dedicated team will be readily accessible to address your feedback within 8 hours and take appropriate measures to rectify any identified issues or facilitate press release takedowns. Ensuring accuracy and reliability are central to our commitment.
