NITDA warns against entering sensitive data into public AI tools

NITDA warned that information entered into public AI platforms may no longer remain under the control of the organisation that supplied it.[1] The agency said service providers could retain or log submitted data, and could use it to train AI models.[1] NITDA also said personally identifiable information could be exposed inadvertently.[1] These are risks the advisory identifies; it does not say that every provider handles data in the same way or that every submission will be retained or used for training.[1]

NITDA said exposure of personal data could constitute a personal data breach and could violate applicable data-protection laws.[1] It also warned that such exposure could have legal consequences.[1] Separately, the agency said disclosing classified, official-use or confidential information to external AI providers could compromise national security.[1] NITDA warned that these disclosures could expose organisations and employees to disciplinary or legal consequences.[1] The advisory therefore treats sensitive-data exposure as both an organisational security concern and a potential compliance issue.[1]

NITDA advised organisations and employees not to enter confidential, classified, official-use or personal data into public AI platforms.[1] For official work, it recommended using only approved AI tools.[1] The warning was issued through CERRT.NG in an advisory posted on X on Saturday.[1] The Sun published its report on the warning on October 4, 2026, at 8:14 p.m. WAT.[1] NITDA had also warned organisations and individuals in May about DeepLoad, an AI-powered malware capable of stealing browser-stored passwords and other sensitive information.[2]