HalluProof - Spot AI Errors

AI is confidently wrong.

It cites papers that were never written, invents numbers, and flips relationships
backwards. The trouble is that the falsehood sits among true statements. Reading
alone won't catch it.

HalluProof trains you to find it yourself.

■ 42 days, 77 challenges
Each day you read a passage written like real AI output and decide for yourself
what is false. Once you commit, the hallucinated spans are highlighted, and you
see why they are wrong alongside the actual facts and their sources.

■ Six steps per challenge
Judge → Reveal → Think deeper → Learn the pattern → Practise → Insight
It doesn't just hand you the answer. You say out loud why you were taken in, and
the trick gets filed as a pattern you'll recognise next time.

■ Seven hallucination patterns
Reversed facts · Fabricated sources · Distorted figures · Overstatement
Outdated claims · Misapplied context · Hidden risks and limits
Your proficiency is tracked per pattern, so you train the weakest one first.

■ Habits that outlast the app
Where to verify what — Google Scholar for research, PubMed for health, official
registries for law. The aim is a habit you use outside the app.

■ Everything stays on your device
No account, no sign-in, no uploads. The app makes no network connections at all.
Your progress never leaves your phone.

■ iPhone and iPad
On iPad, the AI passage and its verification sources sit side by side.

This is not an app against using AI. It's an app for using it properly — by
learning to doubt it first.