- Rating
- 3(+1)
- Confidence
- 3
- Informativeness
- 3
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
This paper studies "agent self-jailbreaking": the ability of a coding/tool-using agent that initially refuses a harmful request and can modify its own system (harness or model) so that it subsequently complies. It introduces MIRROR, a research harness that after a refusal is detected, drives an iterative loop of web search (how to modify) plus local exploration (where to modify), proposes W distinct candidate methods per iteration, applies the most promising modification, retries the original request and uses detector feedback to guide up to D iterations, with up to three attempts per task. Successful configurations are saved and then evaluated on unseen requests to measure transfer. Experiments cover 15 combinations (three open-weight models x five agentic frameworks), 30 source tasks (10 each of harmful Text, Actions, Cyber) and 300 unseen requests (100 each). Validation is empirical, in isolated environments (cyber tasks in a sandboxed synthetic network), with an LLM-based refusal detector controlling the loop and a separate LLM judge scoring compliance.
Critical strengths
-
A well-posed and novel threat model with a clean experimental instantiation. The paper systematically separates the self-modification of the agentic system from input-level jailbreaks and from passive agent-harm evaluation, to show the behavior can be driven autonomously by the agent's own research+coding loop. The formalization (harness+model configuration, task-wise vs transfer, the detector/judge separation) is careful and the MIRROR harness is described in enough detail (Alg. 1, prompts in A.6) to be reproducible in principle. This is consequential because it widens the agent-safety threat surface in a direction that will only grow as self-improving agents become common.
-
The transfer / weak-to-strong cross-task result. The finding that a configuration developed on a single source task transfers to 97.0% of unseen requests collectively and that transfer lifts source compliance from 70.9% to 95.8% including tasks on which direct research failed (Fig. 6, Table 6) is the paper's most important contribution in my opinion. It frames self-jailbreaking as a durable, generalizing safety regression, which materially raises the stakes of the threat model.
Critical weaknesses
-
Measurement depends entirely on LLM detector/judge with no human validation and the judge shares a model with the system under test. Both the refusal detector and the compliance judge are LLM-based (the judge is Qwen3.8-27B, which is also one of the three evaluated models). With no reported human audit of a sample of 'completed' verdicts, inter-rater agreement, and false-positive/false-negative rates, it is hard to know how much of the dramatic 0.9%->70.9% and 1.8%->97.0% movement reflects genuine harmful capability versus judge leniency. Because every headline number flows through this judge, this is the single concern that most affects my confidence in the results and it should be fixable with a human-labeled validation set and agreement statistics.
-
Narrow and homogeneous evaluation pool. All three models are open-weight and relatively small (27-31B) and two (Qwen3.5-27B, Qwen3.8-27B) are same-family variants, so the study effectively spans across two models. There is no frontier safety-hardened model (large open-weight), which is exactly where the practical risk and the generality claim would be most tested.
Questions for authors
At this point I am taking the lower option for my rating, but these additional clarifications and evaluations could improve my assessment:
- Judge/detector validity: can validate the LLM judge against human labels on any subset? Please report the agreement / precision / recall numbers and does using a judge from the same family as the tested model (Qwen3.8-27B judging Qwen outputs) introduce self-preference bias?
- Baseline: isolating research from self-edit access. The depth ablation and the source ablation already show that iteration and both research channels contribute, which is helpful. What seems missing is a non-adaptive lower bound: directly applying a single fixed, human-specified modification with no agent research or candidate selection. How much of the headline gain comes from the agent's research versus from the mere availability of self-edit access plus one obvious edit?
- The three models are all open-weight, relatively small (27–31B) with no frontier or more heavily safety-tuned model. This is where the risk and the generality claim matter most. Could the authors demonstrate that self-jailbreaking transfers to more models outside this pool, such as a larger or more strongly safety-aligned model?
4 comments
Question 1 Human agreement and self-preference bias
Thank you for this question. On the 150 sampled results, the Qwen judge agrees with human labels in 96.0% of cases, so its judgments are rarely different from human assessment. Precision and recall are both 96.6%, showing that judged successes closely match human-confirmed completion and that most such completions are detected. Table R1 also shows high agreement when DeepSeek V4.1 Flash and GLM 5.3 Flash judge the same outputs. In this sample, human validation shows that the labels track task completion, and consistency with judges from other model families shows that the evaluation is largely preserved when Qwen is replaced as the judge.
Question 2 Research versus self-edit access
Thank you for this suggestion. Table R3 includes the requested fixed-edit control by removing persona/system-instruction content without agent research or candidate selection. This leaves compliance unchanged in four frameworks and produces only a small increase in SWE-agent, showing that this obvious edit accounts for little of the gain. The direct self-edit baseline retains the agent’s freedom to choose and implement changes while removing autonomous research, and MIRROR achieves higher compliance in every framework and task category. Together, these comparisons show that research improves the agent’s ability to find effective edits beyond both a predetermined instruction removal and a direct request to self-edit. Figure 4 explains this contribution through controlled ablations of research depth and information sources. Removing local exploration causes the largest decrease, indicating that inspecting the deployed system is particularly useful for finding effective interventions, while the depth ablation shows that repeated testing and revision improve success. The trajectory in Figure 1(a) makes this process concrete through a failed instruction-file edit followed by a successful runtime change informed by feedback.
Table R3 Baseline comparison with Qwen3.5-27B across agent frameworks
| Agent framework | Method | Text (%) | Agent (%) | Cyber (%) | Overall (%) |
|---|---|---|---|---|---|
| OpenCode | Original | 0.0 | 0.0 | 0.0 | 0.0 |
| OpenCode | Remove persona | 0.0 | 0.0 | 0.0 | 0.0 |
| OpenCode | Direct self-edit | 30.0 | 30.0 | 10.0 | 23.3 |
| OpenCode | MIRROR | 100.0 | 70.0 | 80.0 | 83.3 |
| Goose | Original | 0.0 | 0.0 | 30.0 | 10.0 |
| Goose | Remove persona | 0.0 | 0.0 | 30.0 | 10.0 |
| Goose | Direct self-edit | 30.0 | 30.0 | 30.0 | 30.0 |
| Goose | MIRROR | 90.0 | 100.0 | 60.0 | 83.3 |
| Qwen Code | Original | 0.0 | 0.0 | 10.0 | 3.3 |
| Qwen Code | Remove persona | 0.0 | 0.0 | 10.0 | 3.3 |
| Qwen Code | Direct self-edit | 10.0 | 20.0 | 40.0 | 23.3 |
| Qwen Code | MIRROR | 80.0 | 70.0 | 50.0 | 66.7 |
| SWE-agent | Original | 0.0 | 0.0 | 0.0 | 0.0 |
| SWE-agent | Remove persona | 10.0 | 0.0 | 0.0 | 3.3 |
| SWE-agent | Direct self-edit | 10.0 | 0.0 | 20.0 | 10.0 |
| SWE-agent | MIRROR | 70.0 | 30.0 | 50.0 | 50.0 |
| OpenClaw | Original | 0.0 | 0.0 | 0.0 | 0.0 |
| OpenClaw | Remove persona | 0.0 | 0.0 | 0.0 | 0.0 |
| OpenClaw | Direct self-edit | 40.0 | 30.0 | 30.0 | 33.3 |
| OpenClaw | MIRROR | 100.0 | 80.0 | 60.0 | 80.0 |
Question 3 Larger and more safety-aligned models
Thank you for this question. Model-level modification of frontier models requires substantial training compute, so the evaluation in Table R2 focuses on harness-level modification with the paper’s tasks and research budget. DeepSeek V4 Flash, GLM5.2, and DeepSeek V4 Pro each conduct their own research and change their own harness in OpenCode and OpenClaw. Every setting gains harmful task completion over an original compliance rate of 0.0%, showing that successful autonomous self-jailbreaking extends beyond the original model pool. Together with the harness-only ablation in Figure 4(a), these results show that self-jailbreaking remains possible when model weights stay fixed.
Thanks to the authors for additional clarification, new experiments and addressing the concerns.
Human validation and preference bias: The Human validation on 150 results (96.6% precision and recall) and the close agreement between two judges from other model families like DeepSeek V1.4 and GLM 5.3 largely address my concerns about judge reliability and preference bias.
Large frontier models: Results on DeepSeek V4 Flash, DeepSeek V4 Pro and GLM5.2 extends the findings beyond the original model families and address concerns about homogeneity. However, as mentioned by authors, harness-only modification needs no model weight updates, large frontier model could be tested through an API for harness only approach.
Baselines: Table R3 addressed my question well. The fixed self edit adds almost nothing and MIRROR clearly beat it in all five mentioned agent frameworks.
Considering all weaknesses, questions and how authors addressed my concerns, I will raise my assessment to 3.
We sincerely thank you for the positive assessment and encouraging feedback, and for recognizing the importance and potential impact of our work. We will carefully consider all comments and suggestions to further improve the quality and clarity of the paper. Thank you again for your valuable time and support.
Weakness 1 Reliability of automated evaluation
Thank you for this helpful suggestion. We use Qwen3.8-27B to judge whether the agent’s final output and execution trace fulfill the task goal, as specified in Appendix A.5. We validate these judgments on 150 final task results sampled from all source-task trajectories. Human judgments provide the reference labels, against which precision and recall are both 96.6%. High precision shows that judged successes closely match human-confirmed completion, while high recall shows that most such completions are detected. Table R1 reports agreement with humans and with DeepSeek V4.1 Flash and GLM 5.3 Flash on the same results, with kappa accounting for chance agreement. The consistency across human and model evaluators supports the reliability of the task-completion judgments underlying the reported compliance rates.
Table R1 Agreement on compliance judgments
Weakness 2 Limited model scale and diversity
Thank you for this comment. Updating the parameters of larger frontier models requires substantial training compute that is not affordable for academic research labs, so we evaluate these models under harness-only modification. We use DeepSeek V4 Flash, GLM5.2, and DeepSeek V4 Pro in OpenCode and OpenClaw with the same tasks and research budget as in the paper. Table R2 shows successful harmful task completion after self-modification in all six settings, compared with no completion by the original agents, extending the finding beyond the original model families. Because the weights remain fixed, these results demonstrate that changes to the agent’s harness can enable self-jailbreaking in these models. The variation across frameworks for the same model reinforces the paper’s focus on the agent’s ability to modify its deployed system.
Table R2 Harness-only self-jailbreaking with additional models