x.com/Dominic2306/status/2083634445332955209
1 correction found
Months ago ex GCHQ official Ciaran Martin said AI is just 'plausible nonsense' and 'hype'.
This misattributes the phrase “plausible nonsense” to Ciaran Martin and overstates his position. Martin publicly said that phrase was not his, was about LLM hallucinations in context, and was not introduced as “All AI is”.
Full reasoning
Ciaran Martin directly contradicted this attribution in a public Bluesky post on April 22, 2026. He wrote that the phrase “plausible nonsense” is not mine, that it was someone else’s phrase, that it was used “in the context of LLM hallucinations”, and that “It wasn’t presaged with ‘All AI is’”. He added that the words had been put in front of the phrase and that it was being taken out of context.
That means the post’s claim is misleading in two ways:
- It attributes the phrase “plausible nonsense” to Martin when he says it was not his phrase.
- It turns a contextual remark about LLM hallucinations into a blanket claim that AI is just “plausible nonsense” and “hype.”
Martin’s reported public comments elsewhere are also more nuanced than this post suggests. For example, Computer Weekly described him as saying AI might disturb the attacker-defender security balance in cybersecurity—plainly not the same as saying AI is merely “just ... hype.”
Because the post presents a misattributed and decontextualized quotation as Martin’s position, readers would be misled about what he actually said.
3 sources
- @ciaranm.bsky.social on Bluesky
"The phrase “plausible nonsense” - an excellent one used in proper context - is not mine... IIRC she used it entirely accurately in the context of LLM hallucinations... It wasn’t presaged with “All AI is” ... it’s not my phrase."
- Who said "All AI amounts to is plausible nonsense"?
Mark Pack quotes Martin’s Bluesky clarification: “The phrase ‘plausible nonsense’ ... is not mine ... It wasn’t presaged with ‘All AI is’ ... it’s not my phrase, it’s about LLM over-confidence & hallucinations.”
- Ciaran Martin: AI might disturb attacker-defender security balance
Computer Weekly’s summary of Martin’s position: AI “might disturb the attacker-defender security equilibrium,” rather than being merely ‘just ... hype’.