The enhancement addresses a critical challenge facing organizations under India’s DPDP Act: employees using powerful AI tools through personal accounts, creating uncontrolled data flows & leaks that bypass corporate visibility and compliance requirements.
The AI Productivity Paradox
Following the December 2025 Meta acquisition of Manus, enterprises face a new complexity. Manus – which processes over 147 trillion tokens and has created over 80 million virtual computers – operates as an autonomous execution engine capable of browsing the web, writing and deploying code, analyzing data, and producing reports. Unlike ChatGPT’s conversational interface, Manus completes multi-step workflows end-to-end with minimal human intervention.
Govind Rammurthy, CEO & Managing Director, eScan
“The Samsung semiconductor leak demonstrated what happens when employees share sensitive data with ChatGPT,” said Govind Rammurthy, CEO & Managing Director, eScan. “But Manus represents a different category of risk entirely. It’s not just receiving data – it’s autonomously executing tasks that could involve accessing corporate systems, writing code for production environments, or conducting competitive research using proprietary information. Without tenant control, organizations have zero visibility into what employees are asking these agents to do.”
Claude, Anthropic’s AI assistant, has seen explosive enterprise adoption for technical documentation, code review, and strategic analysis. Like ChatGPT, Claude allows users to authenticate through personal email addresses, Google or Microsoft accounts – creating the same data sovereignty challenges that triggered eScan’s initial ChatGPT Tenant Control feature.
How eScan’s Solution Works
eScan Enterprise DLP’s Workspace Tenant Control now monitors authentication attempts to Claude.ai and Manus.im at the endpoint level. When an employee attempts to access either platform using personal credentials or third-party SSO providers (Google, Apple, Microsoft), the DLP system intercepts and blocks the login attempt. Authentication succeeds only when employees use their corporate domain credentials.
The solution maintains productivity while ensuring compliance: employees retain access to AI tools they need for legitimate work, but all interactions occur through accounts the organization can audit, monitor, and control under DPDP requirements.
For Manus specifically, this becomes particularly critical given Meta’s integration plans. The company announced it will embed Manus capabilities into Facebook, Instagram, WhatsApp, and Meta AI. Employees attempting to use Manus through personal Meta accounts – whether standalone or through social platforms – trigger the same tenant control enforcement.
“Organizations told us they blocked ChatGPT entirely after the Samsung leak,” Rammurthy explained. “Then they blocked Claude when developers started using it for code review. Now they’re asking about Manus because employees are using it for market research and competitive analysis. Blocking every AI tool isn’t sustainable – employees will find workarounds. Tenant control lets you enable productivity while maintaining governance.”
The Broader Context
GitHub reported 39 million secrets leaked across its platform in 2024. The March 2025 tj-actions GitHub compromise exposed CI/CD credentials across 23,000 repositories. In February 2025, a Pune resident lost ₹43 lakh to scammers using AI-generated deepfakes. India’s DPDP Act has transformed data leakage from a technical problem into a compliance crisis.
eScan’s Workspace Tenant Control already manages authentication for Google Workspace, Microsoft 365, Dropbox, Atlassian, Slack, Webex, Autodesk, Zoom, WeTransfer, ChatGPT, GitHub, and dozens of other platforms. The addition of Claude and Manus extends this unified governance framework to cover the rapidly expanding AI agent ecosystem.
The enhanced capability is now part of eScan’s Enterprise DLP solution, with support for additional AI platforms planned based on customer requirements. As AI tools proliferate and become embedded in daily workflows, the ability to ensure corporate data sovereignty across these platforms becomes not just a security feature but a fundamental requirement for responsible AI adoption.
