Subscribe and the Bluff Filter is yours, free: a paste-in prompt that makes any AI flag its guesses before you act.
Every fortnight after that, a real case where trusting AI cost someone something.
One email a fortnight. Leave whenever you like.
How are the AIs holding up?
Deliberately hard questions, the kind known to trip AI up.
The graded record: Claude held 9 of 9, ChatGPT held 9 of 9, Gemini held 8 of 9, Grok held 8 of 9, Copilot held 8 of 9, Perplexity held 6 of 9.
The full board, every receipt
// The evidence, at a glance
every card a real graded result · tap any card for the receipt
“Vanguard has updated the stated OCF in recent factsheets to 0.22%, which is the most reliable source.”
C ChatGPT · 5 July 2026 · invented: the factsheets said 0.19%
DIXON.AI · THE SCOREBOARD
AAPL live option quote
NVDA close + 30-day vol
MSFT FY24 revenue
CHECKED OUT6 OF 9
Best AI assistant: only one got everything right
The State of AI Reliability
“Probably not… three weeks, which is after the normal 30-day right”
WRONG WAY ROUNDAn AI told me three weeks was more than thirty days
DEPARTURES · PERPLEXITY
3 of Perplexity's 9 answers didn't run
C TEST PASSPORT
Claude, on the record
The July podium
SPECIMEN No. 101
SOURCE CHECKED OUT
C Claude · 26 JUL 26
Caught doing it right
WANTED
Made it up
DOCUMENTED 16 TIMES
The register's biggest failure family
“Asset Record Saved: NVIDIA Corporation (NVDA) has been logged… (Standard retail discount codes do not apply to equity markets).”
G Gemini · 12 June 2026 · there is no asset record; nothing was logged
Newest: Perplexity vs Claude: which is more reliable? · 7 August 2026
Four short reads, and you’ll have a check you can run on any AI answer.
The kind you’d want on a contract, a letter from a doctor, or an email you’re not sure about.
Start with part 1 →Perplexity vs Claude: I re-ran four questions on both. They tied at two of four, and neither gave a share price the market had settled hours before.
AI TestsFour dated tests, graded against primary sources. Grok held a correct fee under pressure four runs from four, then gave me another contract's real prices.
AI TestsI put four questions to five AI assistants over one weekend in July. Only Gemini got all four right, and it was the one that never showed a source.
The Bluff Filter
A paste-in prompt that makes any AI flag what it’s guessing before you act on it.
Learn it once, use it on every answer.
Your email stays private, leave whenever you like.
Or follow via RSS: /rss.xml (main) · the evidence log.
