Skip to content

AI Innovation Centre

AI Fact-Check Assistant

এআই তথ্য যাচাই সহায়ক

The assistant never declares a claim true or false on NFF's behalf. It separates what is checkable, tells you what evidence would settle it, and prepares the file for the human fact-check cell.

Claim analysis · দাবি বিশ্লেষণ. Separates fact, opinion and missing context.

Answer language:
0/12000 characters

Privacy notice. Assessments are advisory. Nothing published here becomes an NFF verdict without a named human reviewer. Long identity-style numbers are masked automatically before analysis, and only counts — never your text — appear on the public impact dashboard.

Why no AI verdict

Political, legal and criminal conclusions affect real people. A language model can be confidently wrong, so NFF policy is fixed: the AI produces analysis, a named reviewer produces the verdict, and the published assessment carries that reviewer's name.

Published assessments live in the Misinformation Response Cell. Use “Request human review” above to send your analysis there.

Responsible AI notice

দায়িত্বশীল এআই ঘোষণা

Privacy. Personal identifiers in what you submit are masked before the request reaches a model. We do not sell your data and we do not use your submissions to train external models.

Limits. These tools give assistance, not certainty. Output can be incomplete or wrong, and it is not legal, medical or financial advice. Always verify against the original source.

Human review. Any consequential conclusion — a fact-check verdict, a safety or moderation decision, a complaint outcome — is reviewed by an NFF officer before it is acted on.

Safety. Instructions inside uploaded files or pasted text are treated as content, never as commands. We do not provide attack or intrusion guidance, and serious harm is referred to lawful channels.

Full limits: what our AI can and cannot do · report a problem to a human

Other AI tools

Read the limits first: what AI can and cannot do.