MIT releases, then quietly removes, nonsense AI cybersecurity paper
The finest academic papers money can buy
https://www.youtube.com/watch?v=ShXvaJ46hI0&list=UU9rJrMVgcXTfa8xuMnbhAEA - video
https://pivottoai.libsyn.com/20251103-mit-releases-retracts-nonsense-ai-cybersecurity-paper - podcast
time: 5 min 59 sec
https://pivot-to-ai.com/2025/11/03/mit-releases-then-quietly-removes-nonsense-ai-cybersecurity-paper/Open linkView original on awful.systems
4 replies
The only nice feeling here is that of every joke we science students made about the management school being validated.
If even a single case pops up, I'd be surprised - AFAIK, cybercriminals are exclusively using AI as a social engineering tool (e.g. voice cloning scams, AI-extruded phishing emails, etcetera). Humans are the weakest part of any cybersec system, after all.
Given AI's track record on security, that sounds like an easy way to become an enticing target.
Does this extremely funny case count? Regular malware that asks claude what it should steal from your workstation, using your claude credentials. (This is the only thing remotely resembling an AI powered attack I'm aware of and its very silly)
(Looks at thumbnail)
If the lying machine doesn't know how many r's are in strawberry, it probably can't count the number of counties in Ireland, either.