The National Information Technology Development Agency has warned its workers against entering classified government information, personal data and other confidential materials into public artificial intelligence platforms.
NITDA issued the warning in an advisory on the safe use of AI large language models on Monday, citing the growing use of tools such as ChatGPT, Claude, Gemini and Copilot for research, drafting and other work-related tasks.
The agency said the use of such platforms could expose sensitive information because data entered into public AI systems might be retained, logged or used to train the providers’ models.
NITDA warned that exposing personally identifiable information through public AI platforms could amount to a personal data breach and violate applicable data protection laws, with possible legal consequences.
It also cautioned that uploading classified, official-use or confidential government information to external AI providers could compromise national security and public trust.
The agency advised staff to use only AI tools approved by their organisations for official work and to remove, anonymise or pseudonymise personal and sensitive information before using AI platforms.
NITDA further urged users to check the privacy and data-handling terms of AI platforms before using them and warned staff against uploading internal documents without specific authorisation.
The agency said employees should also comply with existing AI, information security and data protection policies when using the technology for official duties.
