Pith. sign in

REVIEW 19 cited by

Image Hijacks: Adversarial Images can Control Generative Models at Runtime

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2309.00236 v4 pith:C4B2FMJV submitted 2023-09-01 cs.LG cs.CLcs.CR

Image Hijacks: Adversarial Images can Control Generative Models at Runtime

classification cs.LG cs.CLcs.CR
keywords hijacksimagebehaviourmatchingpromptadversarialattackattacks
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original 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 located in Rome') using a generic, off-the-shelf dataset unrelated to our choice of prompt. We use Behaviour Matching to craft hijacks for four types of attack, forcing VLMs to generate outputs of the adversary's choice, leak information from their context window, override their safety training, and believe false statements. We study these attacks against LLaVA, a state-of-the-art VLM based on CLIP and LLaMA-2, and find that all attack types achieve a success rate of over 80%. Moreover, our attacks are automated and require only small image perturbations.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 19 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Fine-tuning Multi-modal LLMs with ART: Art-based Reinforcement Training

    cs.LG 2026-06 unverdicted novelty 7.0

    ART optimizes visual pixel inputs to frozen MLLMs to achieve LoRA-competitive accuracy on math and structured tool-use benchmarks without modifying computational graphs.

  2. VisualLeakBench: Reproducible Action-Boundary Propagation Failures in Vision-Language Agents

    cs.CV 2026-05 unverdicted novelty 7.0

    VisualLeakBench shows vision-language agents propagate PII and unsafe text from images to tool arguments at 78.8% and 85.5% baseline rates, with defensive prompts reducing PII leakage to 2% mainly by suppressing tool ...

  3. Catching the Infection Before It Spreads: Foresight-Guided Defense in Multi-Agent Systems

    cs.AI 2026-05 unverdicted novelty 7.0

    A foresight-based local purification method using multi-persona simulations and recursive diagnosis reduces infectious jailbreak spread in multi-agent systems from over 95% to below 5.47% while matching benign perform...

  4. Gaslight, Gatekeep, V1-V3: Early Visual Cortex Alignment Shields Vision-Language Models from Sycophantic Manipulation

    cs.CV 2026-04 conditional novelty 7.0

    Alignment of vision-language models with human V1-V3 early visual cortex negatively predicts resistance to sycophantic gaslighting attacks.

  5. Visual Adversarial Attack on Vision-Language Models for Autonomous Driving

    cs.CV 2024-11 unverdicted novelty 7.0

    ADvLM is the first visual adversarial attack framework for VLMs in autonomous driving, using semantic-invariant induction via LLM-generated prompt libraries and scenario-associated attention-based enhancement to achie...

  6. On Adversarial Vulnerability of Vision-Language Models through the Lens of Intermediate Spectral Subspaces

    cs.LG 2026-07 conditional novelty 6.0

    Aligning adversarial perturbations with the near-null singular directions of intermediate linear layers in transformer VLMs yields stronger attacks than existing feature- and output-space methods.

  7. Exploring Adversarial Robustness and Safety Alignment in Multilingual Multi-Modal Large Language Models

    cs.CL 2026-06 unverdicted novelty 6.0

    Adversarial images transfer across languages in MLLMs while apparent safety in weaker languages stems from comprehension and visual-grounding failures rather than genuine alignment.

  8. Benign Inputs, Harmful Outputs: Cross-Modal Jailbreaking via Distributed Semantic Recomposition

    cs.CR 2026-06 unverdicted novelty 6.0

    DSR decomposes harmful intents into benign textual and visual primitives that MLLMs fuse into harmful outputs, achieving high attack success with low input toxicity.

  9. Guaranteed Jailbreaking Defense via Disrupt-and-Rectify Smoothing

    cs.CR 2026-05 unverdicted novelty 6.0

    DR-Smoothing introduces a disrupt-then-rectify prompt processing scheme into smoothing defenses, delivering tight theoretical bounds on success probability against both token- and prompt-level jailbreaks.

  10. Catching the Infection Before It Spreads: Foresight-Guided Defense in Multi-Agent Systems

    cs.AI 2026-05 unverdicted novelty 6.0

    FLP uses multi-persona foresight simulation to detect infections via response diversity and applies local purification to reduce maximum cumulative infection rates in multi-agent systems from over 95% to below 5.47%.

  11. Catching the Infection Before It Spreads: Foresight-Guided Defense in Multi-Agent Systems

    cs.AI 2026-05 unverdicted novelty 6.0

    A foresight-based local purification method simulates future agent interactions, detects infections via response diversity across personas, and applies targeted rollback or recursive diagnosis to cut maximum infection...

  12. VisInject: Disruption != Injection -- A Dual-Dimension Evaluation of Universal Adversarial Attacks on Vision-Language Models

    cs.CR 2026-05 conditional novelty 6.0

    Universal adversarial attacks cause output perturbation 90 times more often than precise target injection in VLMs, with only 2 verbatim successes out of 6615 tests.

  13. RedDiffuser: Auditing Multimodal Safety Failures in Vision-Language Models via Reinforced Diffusion

    cs.CV 2025-03 unverdicted novelty 6.0

    RedDiffuser is a reinforced diffusion framework that generates adversarial visual contexts to audit and expose widespread multimodal safety failures in VLMs, increasing unsafe response rates by up to 10.69% on LLaVA w...

  14. A StrongREJECT for Empty Jailbreaks

    cs.LG 2024-02 conditional novelty 6.0

    StrongREJECT provides a standardized benchmark and evaluator for jailbreak attacks that aligns better with human judgments than prior methods and reveals that successful jailbreaks often reduce model capabilities.

  15. MIRAGE: Stealthy Visual Prompt Injection for Vulnerability Detection in Web Agents

    cs.CV 2026-06 unverdicted novelty 5.0

    MIRAGE creates perceptually benign adversarial images using diffusion and curvature-aware optimization to enable targeted prompt injection attacks on web agents like SeeAct and OpenClaw within attacker-controlled boundaries.

  16. Laundering AI Authority with Adversarial Examples

    cs.CR 2026-05 unverdicted novelty 5.0

    Adversarial examples enable AI authority laundering by causing production VLMs to give authoritative but wrong responses on subtly perturbed images, with success rates of 22-100% using decade-old attack methods.

  17. Surviving the Unseen: Predictive Defense for Novel Multi-Turn Multimodal Attacks

    cs.CR 2026-05 unverdicted novelty 4.0

    Proposes the TRIAD framework that treats multi-turn multimodal attacks as continuous trajectories and uses structural anomaly detection, regularized Mahalanobis distance, topological acceleration, and a time-varying C...

  18. AI Safety Landscape for Large Language Models: Taxonomy, State-of-the-art, and Future Directions

    cs.AI 2024-08 unverdicted novelty 4.0

    The paper introduces a taxonomy of AI safety for LLMs organized into Trustworthy AI, Responsible AI, and Safe AI perspectives, accompanied by a review of state-of-the-art methods, challenges, and future directions.

  19. Safety at Scale: A Comprehensive Survey of Large Model and Agent Safety

    cs.CR 2025-02 unverdicted novelty 2.0

    A comprehensive survey that taxonomizes safety threats to large models and agents, reviews defenses and benchmarks, and outlines open challenges.