all InfoSec news
Image Hijacks: Adversarial Images can Control Generative Models at Runtime
April 24, 2024, 4:11 a.m. | Luke Bailey, Euan Ong, Stuart Russell, Scott Emmons
cs.CR updates on arXiv.org arxiv.org
Abstract: Are foundation models secure against malicious actors? In this work, we focus on the image input to a vision-language model (VLM). We discover image hijacks, adversarial images that control the behaviour of VLMs at inference time, and introduce the general Behaviour Matching algorithm for training image hijacks. From this, we derive the Prompt Matching method, allowing us to train hijacks matching the behaviour of an arbitrary user-defined text prompt (e.g. 'the Eiffel Tower is now …
adversarial algorithm arxiv can control cs.cl cs.cr cs.lg discover focus foundation foundation models general generative generative models image images input language malicious malicious actors runtime training work
More from arxiv.org / cs.CR updates on arXiv.org
Jobs in InfoSec / Cybersecurity
Information Security Engineers
@ D. E. Shaw Research | New York City
Technology Security Analyst
@ Halton Region | Oakville, Ontario, Canada
Senior Cyber Security Analyst
@ Valley Water | San Jose, CA
Consultant Sécurité SI Gouvernance - Risques - Conformité H/F - Strasbourg
@ Hifield | Strasbourg, France
Lead Security Specialist
@ KBR, Inc. | USA, Dallas, 8121 Lemmon Ave, Suite 550, Texas
Consultant SOC / CERT H/F
@ Hifield | Sèvres, France