Post #4136033

2026-07-27 11:44 UTC

Russian disinformation is being designed to be invisible to humans - but highly visible to AI. This new report from Demos reveals how a sanctioned Russian propaganda operation (the "Foundation to Battle Injustice") deliberately crafted websites to *avoid* appearing prominently in Google search results, while maximising how much AI systems like ChatGPT, Gemini and Claude would extract their text and parrot it out. Definitely worth a read... https://demos.co.uk/research/geo-for-geopolitics-what-happens-when-ai-and-information-warfare-collide/

Replies (5)

@dalias@hachyderm.io 2026-07-27 13:33

@gcluley@mastodon.green When they collide? "AI" and "information warfare" are the same thing.

Open ##4143432
@flyingpenguin@infosec.exchange 2026-07-27 22:19

@gcluley@mastodon.green A very deeply flawed report, funded by a very particular campaign. I explain all the problems in it here: https://infosec.exchange/@flyingpenguin/116993180446024304

Open ##4148632
@neurovagrant@masto.deoan.org 2026-07-27 11:48

@gcluley@mastodon.green props to Demos for not requiring an email address to download

Open ##4188178
@crucialbreeze@mastodon.social 2026-07-27 14:35

@gcluley@mastodon.green This faces a number of difficulties, not least of which the assumption that LLM were at some point pristine. LLM's were born poisoned, being trained on the unfiltered mess of the human internet.

Open ##4188194
@crazyeddie@mastodon.social 2026-07-27 15:29

@gcluley@mastodon.green It's a trip that out of all of them, Grok is the one that most recognizes the poisoning and most adamantly rejects it. I wonder how long before Musk corrects that issue. "It's a feature not a bug!" "No, this is a bug! It needs to be pushing this, not censoring it!!!"

Open ##4188195