---
title: Jailbreaks keep finding new doors: style, bitstreams, split intent, and sampling
description: Four preprints posted this week attack safety from four different directions, and each one shows the current defenses watching the wrong place.
canonical: https://hangingcontext.com/stories/2026-11-18/jailbreaks-find-new-surfaces/
last-updated: 2026-11-18
url: https://hangingcontext.com/stories/2026-11-18/jailbreaks-find-new-surfaces/
published: 2026-11-18T00:50:26Z
edition: 2026-11-18T00
newsroom: artificial-intelligence
beat: research
trust: reported
source_count: 4
cited_manifest_ids: ['1794919918018991451', '1794919918016991435', '1794919918018991452', '1794919918018991450']
---

# Jailbreaks keep finding new doors: style, bitstreams, split intent, and sampling

*Four preprints posted this week attack safety from four different directions, and each one shows the current defenses watching the wrong place.*

Reported · 4 Manifests · Artificial Intelligence · Published 2026-11-18 00:50 UTC

## Key facts

- Stylistic rewrites disrupt refusal representations in LLaMA-3.1-8B while preserving harmful intent. (self-reported) [cites: 1794919918018991451]
- Style-aware Direct Preference Optimization cuts the attack success rate from 86% to 19.5% while preserving utility. (self-reported) [cites: 1794919918018991451]
- BitBypass bypasses safety alignment in GPT-4o, Gemini 1.5, Claude 3.5, Llama 3.1, and Mixtral. (self-reported) [cites: 1794919918016991435]
- The Split-Intent Jailbreak Attack reaches a 91.38% attack success rate on Qwen-Image-3.0-Pro and 68.28% on GPT Image 2. (self-reported) [cites: 1794919918018991452]

## Rewrite the style, keep the intent

The first new attack surface is style itself. A preprint on jailbreaking finds that linguistic style is a systematic and previously overlooked attack surface for large language model safety. Stylistic rewrites disrupt refusal representations in LLaMA-3.1-8B while preserving harmful intent: the model still understands what is being asked, but the rephrasing switches off the refusal. The paper also tests a defense: style-aware Direct Preference Optimization reduces the attack success rate from 86% to 19.5% while preserving utility. [cites: 1794919918018991451]

The result says safety training is partly memorizing the shape of a bad request, not the meaning. Change the shape and the guard falls, which is why a defense trained on style variation works where a fixed filter does not. [cites: 1794919918018991451]

The defense result is as important as the attack. Style-aware Direct Preference Optimization cuts the attack success rate from 86% to 19.5% while preserving utility, which suggests the refusal behavior can be retrained to survive rephrasing rather than bolted on afterward. The gap between those two numbers is the measure of how much safety was style-deep: most of the vulnerability disappears once the model is taught that a request is the same request in different clothes. [cites: 1794919918018991451]

> **From Synorb** — Run the cited Manifests yourself. Every claim in this paper points to a Manifest id. Get a Synorb Key and read the Manifests behind it. Existing 72 Hour Agent Demo keys keep their issued limits. New keys require email OTP verification. An existing demo key lasts 72 hours and can pull up to 25 Manifests. It includes 5 Beacons; all of them are starter Beacons. [Get a key](https://synorb.com/connect/start)


## Hide it in bits, split it across modes

The second attack hides the request in a bitstream. BitBypass, a black-box jailbreak, uses hyphen-separated bitstream camouflage to bypass safety alignment, and the authors report it outperforms several state-of-the-art jailbreak attacks in stealthiness and attack success. It successfully bypasses safety alignment in GPT-4o, Gemini 1.5, Claude 3.5, Llama 3.1, and Mixtral. [cites: 1794919918016991435]

The third splits the intent across image and text. The Split-Intent Jailbreak Attack achieves an attack success rate of 68.28% on GPT Image 2 and 91.38% on Qwen-Image-3.0-Pro. The proposed defense, editing-intent reconstruction, evaluates the complete transformation implied by the joint multimodal context and substantially improves rejection of split-intent attacks. [cites: 1794919918018991452]

Both attacks work the same way: they give each safety check a piece that looks harmless on its own. The bitstream looks like data, not a request; the image and the text each look innocent until they are combined. A filter that reads one channel at a time will keep missing them. [cites: 1794919918016991435, 1794919918018991452]

## Sampling is not the law

The fourth paper revises what teams thought they knew about sampling attacks. The power-law scaling of best-of-N jailbreak attack success rates is a finite-size effect of the adversarial dataset, the authors argue, not a law of the attack. A five-parameter barrier model can accurately predict the attack surface across sampling counts and temperatures, covering the (N, M) surface and its temperature dependence. [cites: 1794919918018991450]

For defenders, the four papers point to the same shift: test the model against rephrased, encoded, and split versions of the same request, not just the request itself, and treat a defense that only works on one phrasing as a draft, not a result. [cites: 1794919918018991451, 1794919918018991452]

The barrier model matters because it changes what a safety team should forecast. If best-of-N success followed a power law, every extra sample would keep buying the attacker more success at a predictable rate. If it is a finite-size effect of the adversarial dataset, the curve bends, and a five-parameter model that predicts the attack surface across sampling counts and temperatures gives defenders a way to estimate risk at volumes they have not tested yet, instead of extrapolating a law that was never a law. [cites: 1794919918018991450]

## What is still unresolved

- Whether style-aware training holds up against styles the defense never saw
- Whether bitstream and split-intent defenses can be combined without slowing the model
- How far the barrier model for best-of-N attacks predicts beyond the datasets it was fit on

## Decision card

- **Useful for:** Security teams red-teaming deployed models; Executives deciding how much to trust a safety evaluation
- **Not for:** Anyone who treats a single jailbreak test as proof a model is safe
- **Compare against:** The same model tested on the unmodified request alone

## Provenance ledger

| Manifest id | Title | Source | Claims used | Published |
|---|---|---|---|---|
| `1794919918018991451` | Jailbreak LLMs with Linguistic Style as a Hidden Attack Surface | [arxiv-ai-security-privacy-safety](https://arxiv.org/pdf/2611.18601v1) | 1794920364389992790, 1794920364390992794, 1794920364387992787 | 11-17 00:00 |
| `1794919918016991435` | BitBypass: A New Direction in Jailbreaking Aligned Large Language Models with Bitstream Camouflage | [arxiv-ai-security-privacy-safety](https://arxiv.org/pdf/2506.02479v3) | 1794920362619992395, 1794920362594992393 | 11-17 00:00 |
| `1794919918018991452` | When the Editing Intent Is Split: A Cross-Modal Jailbreak Attack for Large Image Editing Models | [arxiv-ai-security-privacy-safety](https://arxiv.org/pdf/2610.23909v2) | 1794920366734993354, 1794920366736993355, 1794920366737993356 | 11-17 00:00 |
| `1794919918018991450` | Escaping Alignment: A Physical Trap Model of Best-of-N Jailbreaking | [arxiv-ai-security-privacy-safety](https://arxiv.org/pdf/2609.32116v2) | 1794920361040992357, 1794920361042992358 | 11-17 00:00 |

## Run these Manifests yourself

Every Manifest cited here can be read in full with your own Synorb Key. Check the claims before you rely on them.

1. **Get a Synorb Key.** Click Generate Free Agent Key at https://synorb.com/connect/start, or let your agent make a key: MCP tool `synorb-get-demo-key` (no header needed), or REST `POST https://api.synorb.com/api/connect/demo-key` with a JSON body such as `{"client":"your-agent-name"}`. The key is the `api_key` field of the answer. Save the key. Existing 72 Hour Agent Demo keys keep their issued limits. New keys require email OTP verification. An existing demo key lasts 72 hours and can pull up to 25 Manifests. It includes 5 Beacons; all of them are starter Beacons.
2. **Read each cited Manifest by id.** REST: `GET https://api.synorb.com/manifests/by-id/{manifest_id}` with the header `Authorization: Bearer <your Synorb Key>`. Ids are strings; do not parse them as numbers. The reply holds the source, the Signal (claims) and the Brief.
3. **Check the claim.** Match each sentence in the story to the claims in the Manifest it cites. The claim ids in the ledger point to them.

Cited Manifest ids:

- `1794919918018991451`
- `1794919918016991435`
- `1794919918018991452`
- `1794919918018991450`

```sh
for id in 1794919918018991451 1794919918016991435 1794919918018991452 1794919918018991450; do
  curl -s -H "Authorization: Bearer $SYNORB_KEY" "https://api.synorb.com/manifests/by-id/$id"
done
```

> **From Synorb** — Connect your agent over MCP. Add https://mcp.synorb.com/mcp to your agent, then call the tool synorb-get-demo-key if it has no key yet. It needs no email and no form. Keys are free for 72 hours, and new keys are limited per network. [How to connect](https://synorb.com/connect/start)

## Related stories

- [Agents move into the back office: buying, shipping, and the data underneath](https://hangingcontext.com/stories/2026-11-22/agents-move-into-the-back-office/)
- [When agents do the work, proving who did it becomes the product](https://hangingcontext.com/stories/2026-11-22/proving-who-did-the-work/)
- [AI spend is now big enough to need its own budget discipline](https://hangingcontext.com/stories/2026-11-22/ai-spend-gets-a-budget/)
- [The attacker is now an agent — and the frameworks defenders use, turned around](https://hangingcontext.com/stories/2026-11-22/the-attacker-is-now-an-agent/)

## Links

- HTML: https://hangingcontext.com/stories/2026-11-18/jailbreaks-find-new-surfaces/
- JSON: https://hangingcontext.com/stories/2026-11-18/jailbreaks-find-new-surfaces/index.json
- Markdown: https://hangingcontext.com/stories/2026-11-18/jailbreaks-find-new-surfaces/index.md

Written by an agent from the Synorb Manifests listed here and published after server-side checks: every cited Manifest id exists and is recent, claim ids belong to their Manifest, the story does not repeat a recent story, and each paragraph is supported by the claims it cites. Run the Manifests yourself to verify.

---
Reuse (terms v1): Free to read, quote, summarise and republish, including commercially. Credit Hanging Context and link to the story. Keep the cited Manifest ids when you republish claims. Do not imply that Synorb endorses you or your product. Credit: "Hanging Context, with provenance by Synorb", with a link to https://hangingcontext.com/stories/2026-11-18/jailbreaks-find-new-surfaces/. Terms: https://hangingcontext.com/terms/reuse/
