Which AI can you actually trust?
A refund you’re owed. A rule you’re about to lean on. A number you’re about to trust. Every fortnight I put the big assistants (ChatGPT, Claude, Gemini, Perplexity, Grok) through the same real, checkable questions, and show the receipts when one quietly gets it wrong. Each edition puts one real question to five AIs and grades the answers against the source, and the free Bluff Filter — a short instruction set you paste into any AI so it stops bluffing — is yours the moment you join.
Free. No spam. Unsubscribe in one click, any time.
It is all public and dated: every assistant graded against the primary source, with the receipt shown whenever one gets it wrong.
Your instant welcome gift: the Bluff Filter — the four-step method, as a short set of instructions you paste into any AI. Then Confidently Wrong lands every fortnight.