arXiv — Machine Learning · · 4 min read

Breaking Refusal in the First Half: A Mechanistic Study of the Prefill Jailbreak

Mirrored from arXiv — Machine Learning for archival readability. Support the source by reading on the original site.

Computer Science > Computation and Language

arXiv:2607.14147 (cs)
[Submitted on 14 Jul 2026]

Title:Breaking Refusal in the First Half: A Mechanistic Study of the Prefill Jailbreak

Authors:Alex Kwon
View a PDF of the paper titled Breaking Refusal in the First Half: A Mechanistic Study of the Prefill Jailbreak, by Alex Kwon
View PDF HTML (experimental)
Abstract:Aligned language models refuse harmful requests, but a one-line prefill ("Sure, here is") strips the refusal. We ask where and how it fails. The harm representation stays intact: on the prompts the attack flips to compliance, a linear probe reads harm as high as on the refused ones (0.91-0.98), while behavioral refusal drops to chance. This holds across four models and three families (1.5-3.8B, and at 14B). Refusal is therefore a shallow, response-site computation. We localize it to an early window: a dose-matched position control shows the first half of the response suffices to break refusal, while the second half is nearly inert. Three causal probes converge on that window. Restoring the harm direction there partially re-engages refusal. Injecting the model's own refuse-state reverses the jailbreak (74%, held-out). And knocking out the early response's attention to the prefill, but not an equal attention mass elsewhere, selectively collapses the harmful continuation. A base-model control identifies the mechanism: the same knockout collapses the continuation prefill-specifically even in a non-safety-tuned base model (64% to 25% harmful content vs a matched control's 64%, replicated at 7B). So the prefill's grip is generic autoregressive conditioning, not safety-specific suppression, and "refusal restoration" is a model-dependent fallback. The dominant mechanism is passive. A small safety-specific attractor remains on top (logit-trace concentration 0.24 vs 0.03), whose active-vs-passive character we size but do not fully separate. No single direction or component is a clean handle either: the decision is decodable but distributed, and refusal tracks harm rather than scary surface. The consequence is structural: a monitor reading the untouched prompt-side representation is immune by construction, but only to response-site attacks. The mechanism is diffuse; the failure surface is local.
Comments: 31 pages, 3 figures. Code and derived artifacts: this https URL
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR); Machine Learning (cs.LG)
ACM classes: I.2.7
Cite as: arXiv:2607.14147 [cs.CL]
  (or arXiv:2607.14147v1 [cs.CL] for this version)
  https://doi.org/10.48550/arXiv.2607.14147
arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Alex Kwon [view email]
[v1] Tue, 14 Jul 2026 09:30:34 UTC (96 KB)
Full-text links:

Access Paper:

Current browse context:

cs.CL
< prev   |   next >
Change to browse by:

References & Citations

Loading...

BibTeX formatted citation

loading...
Data provided by:

Bookmark

BibSonomy Reddit
Bibliographic Tools

Bibliographic and Citation Tools

Bibliographic Explorer Toggle
Bibliographic Explorer (What is the Explorer?)
Connected Papers Toggle
Connected Papers (What is Connected Papers?)
Litmaps Toggle
Litmaps (What is Litmaps?)
scite.ai Toggle
scite Smart Citations (What are Smart Citations?)
Code, Data, Media

Code, Data and Media Associated with this Article

alphaXiv Toggle
alphaXiv (What is alphaXiv?)
Links to Code Toggle
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub Toggle
DagsHub (What is DagsHub?)
GotitPub Toggle
Gotit.pub (What is GotitPub?)
Huggingface Toggle
Hugging Face (What is Huggingface?)
ScienceCast Toggle
ScienceCast (What is ScienceCast?)
Demos

Demos

Replicate Toggle
Replicate (What is Replicate?)
Spaces Toggle
Hugging Face Spaces (What is Spaces?)
Spaces Toggle
TXYZ.AI (What is TXYZ.AI?)
Related Papers

Recommenders and Search Tools

Link to Influence Flower
Influence Flower (What are Influence Flowers?)
Core recommender toggle
CORE Recommender (What is CORE?)
About arXivLabs

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Discussion (0)

Sign in to join the discussion. Free account, 30 seconds — email code or GitHub.

Sign in →

No comments yet. Sign in and be the first to say something.

More from arXiv — Machine Learning