Data Security Council Launches India-Specific Guide on Agentic AI Liability

Understanding Agentic AI Liability in India

The Data Security Council of India (DSCI) has published a new industry guide focused on liability frameworks for agentic artificial intelligence. As autonomous AI systems become more prevalent in corporate environments, the guidance aims to help organizations determine legal and operational responsibility when systems deviate from expected tasks or encounter errors.

According to DSCI CEO Vinayak Godse, organizations require practical instruments and clearly defined decision spaces. The newly released report does not introduce entirely new regulatory standards; instead, it interprets how existing binding doctrines apply to multi-vendor and autonomous AI architectures.

Key Highlights of the DSCI Guidance

• Mapping the Value Chain: The report outlines how responsibilities are distributed across various stages of AI deployment.
• Hosting Architecture Impact: The guidance notes that hosting choices play a significant role in determining where liability rests. Architectural control and ownership can heavily influence whether legal responsibility remains centralized or distributed.
• Allocation Across Stakeholders: The framework identifies multiple groups involved in the AI lifecycle, including model providers, tool and framework developers, product companies, integrators, service providers, and user organizations.
• Avoiding Binary Frameworks: The report warns that viewing liability strictly as an either-or choice between end users and upstream players is legally flawed and commercially impractical.

Practical Implications and Verification

The guidance emphasizes that enterprises adopt AI through multiple pathways, such as browsers, co-pilots, embedded enterprise software, and independent models. Each entry route presents distinct liability considerations that require careful scrutiny to prevent hidden exposures.

Industry stakeholders and legal professionals can review the complete framework directly through the official publications of the Data Security Council of India to assess its applicability to specific operational models and compliance strategies.