MacMusic  |  PcMusic  |  440 Software  |  440 Forums  |  440TV  |  Zicos
realizing
Search

How 'sleeper agent' AI assistants can sabotage your code without you realizing

Tuesday January 16, 2024. 10:30 PM , from TheRegister
Today's safety guardrails won't catch these backdoors, study warns
Analysis AI biz Anthropic has published research showing that large language models (LLMs) can be subverted in a way that safety training doesn't currently address.…
Read more at TheRegister
https://go.theregister.com/feed/www.theregister.com/2024/01/16/poisoned_ai_models/

Related News

realizing If you use AI to teach you how to code, remember you still need to think for yourself TheRegisterJan 27
safety Apple's Large Language Model Shows Up in New iOS Code SlashdotJan 27
how 80 Years Later, GCHQ Releases New Images of Nazi Code-Breaking Computer SlashdotJan 19
code [$] Improved code generation in the CPython JIT LWN.netJan 18
without Do users write more insecure code with AI assistants? OS NewsJan 18
'sleeper Behind the Screens: what the computer code seen in movies and TV shows actually does BoingBoingJan 16
sabotage Balancing risk and reward with low-code and no-code platforms [Q&A] BetaNewsJan 15
agent' Your Medical Data Is Code Blue Wired: Tech.Jan 12
assistants New year, new bug – rivalry between devs led to a deep-code disaster TheRegisterJan 8
realizing Enjoy greater productivity with Setapp, now just $73 with this code PC WorldJan 6
safety Code archaeologist digs up oldest known ancestor of MS-DOS TheRegisterJan 5
how Since the Demise of Atom, 'Pulsar' Offers an Alternative Code Editor SlashdotJan 1
code GitHub Makes Copilot Chat Generally Available, Letting Devs Ask Questions About Code SlashdotDec 30
without Code.org Sues WhiteHat Jr. For $3 Million SlashdotDec 26
'sleeper GTA 5 Source Code Reportedly Leaked Online a Year After Rockstar Hack SlashdotDec 26
sabotage Microsoft President Brad Smith Quietly Leaves Board of Nonprofit Code.org SlashdotDec 23
agent' Hundreds of thousands of dollars in crypto stolen after Ledger code poisoned TheRegisterDec 16
assistants Intelligence Researchers To Study Computer Code for Clues To Hackers' Identities SlashdotDec 15
realizing Ex-Wirecard COO Suspected as Decade-Long Russian Agent SlashdotDec 15
safety Cloud Engineer Gets 2 Years For Wiping Ex-Employer's Code Repos SlashdotDec 12
News copyright owned by their original publishers | Copyright © 2004 - 2024 Zicos / 440Network
Current Date
May, Fri 17 - 09:01 CEST