Statement on the Use of AI
Version 2.0 | June 2026
Supersedes Version 1.0, June 2024
Takshashila Institution first published its principles for AI use in June 2024. In the two years since, AI tools have become substantially more capable and are now embedded in how we conduct research, deliver education programmes, create simulation exercises, and communicate our work. This updated statement reflects that reality.
Takshashila actively uses AI tools across its research, education, and communications work. Our mission is to promote India’s national interest through rigorous, independent analysis and education. As a non-profit institution working to transform India, it is our moral responsibility to produce high-quality policy outputs in the most efficient manner. We commit to being the frontier users of any kind of technology in service of that mission.
This policy applies to all research outputs, policy briefs, educational materials, simulation exercises, podcasts, newsletters, and any other work produced by Takshashila. The following principles govern how we use AI.
Accountability
Authors, educators, and content creators are responsible for all work they produce, regardless of what tools assisted in its creation. AI cannot be considered an author or co-creator. Every person producing work at Takshashila is accountable for its compliance with standards of originality, accuracy, and attribution to primary sources.
Transparency
All use of AI in producing Takshashila’s outputs will be disclosed.
Samples of how disclosures can be written:
- This essay was made with the assistance of AI.
- The data visualisation in this piece was made with the assistance of AI.
- This simulation scenario was developed with the assistance of AI.
- The analysis in this paper was supported by AI.
Attribution
Staff and faculty will not use AI to substitute for their own analysis. The role of AI is to improve the quality and productivity of research and education, reduce time spent on searching for leads, structure arguments, brainstorm approaches, or iterate on drafts. Authors and educators are expected to arrive at independent conclusions. All original sources must be correctly cited. All information and sources generated through AI will be independently verified.
Integrity Protocol
Takshashila maintains an institutional integrity protocol for all AI-assisted work, covering three areas:
Review: Where automated output is produced, it is declared as such. Content published without editorial judgment by a named Takshashila researcher will be explicitly declared.
Declaration: All published and educational outputs disclose the nature of AI assistance, in line with the Transparency section above.
Training: Takshashila trains its researchers, faculty, and staff to use AI tools with critical judgment, with awareness of failure modes, limitations, and the institution’s standards of intellectual integrity.
Audit Trail
Takshashila recommends as a best practice that researchers and faculty maintain a record of prompts used as part of a project. This record does not need to be published, but may be used to audit the claims of a document if required.
Reporting Errors
Any person — whether a member of the Takshashila community or an external reader — who identifies an error, inaccuracy, or integrity concern in Takshashila’s published work may report it to the Ombudsman. The Ombudsman is empowered by the Board to review complaints and ensure compliance with Takshashila’s Code of Conduct and institutional standards.