Exposing the Systematic Vulnerability of Open-Weight Models to Prefill Attacks

TL;DR

This paper systematically investigates prefill attacks on open-weight language models, revealing a widespread vulnerability that poses significant risks and highlights the need for improved defenses.

Contribution

It is the first large-scale empirical study of prefill attacks, evaluating over 20 strategies across multiple open-weight models, demonstrating their effectiveness and exposing a critical security gap.

Findings

Prefill attacks are highly effective against all tested models.

Large reasoning models show some robustness but remain vulnerable to tailored attacks.

The study emphasizes the urgent need for defenses against prefill vulnerabilities.

Abstract

As the capabilities of large language models continue to advance, so does their potential for misuse. While closed-source models typically rely on external defenses, open-weight models must primarily depend on internal safeguards to mitigate harmful behavior. Prior red-teaming research has largely focused on input-based jailbreaking and parameter-level manipulations. However, open-weight models also natively support prefilling, which allows an attacker to predefine initial response tokens before generation begins. Despite its potential, this attack vector has received little systematic attention. We present the largest empirical study to date of prefill attacks, evaluating over 20 existing and novel strategies across multiple model families and state-of-the-art open-weight models. Our results show that prefill attacks are consistently effective against all major contemporary open-weight models, revealing a critical and previously underexplored vulnerability with significant implications for deployment. While certain large reasoning models exhibit some robustness against generic prefilling, they remain vulnerable to tailored, model-specific strategies. Our findings underscore the urgent need for model developers to prioritize defenses against prefill attacks in open-weight LLMs.