T-004 · Verify outputs

Confident AI can still be wrong

AI tools generate fluent text, and fluent is not the same as correct. A tool can state a wrong fact in exactly the same tone as a right one.

What to do

  1. Treat the first answer as a draft, not a fact.
  2. Highlight anything specific: names, dates, numbers, quotes, laws.
  3. Confirm each highlighted item with a source you trust.
  4. If you cannot verify it, leave it out.

An example

An AI draft says a supplier "was founded in 2009." You check the supplier's own website. It says 2011. The draft was confidently wrong.

Label copy

For small labels: Confident AI can still be wrong.

For water-bottle labels: AI sounds sure even when it's guessing. Check names, numbers and sources before you use them.

Sources and further reading

Sources are being added by our editor before this tip is finalized. Verify claims against the tool provider's current terms and your own policies.

By Black Rainb0w. Draft, pending human review.