# Prompt Injection Security Evaluation on Professional Domains Corpus Source: [Prompt Injection Security Evaluation on Professional Domains Corpus]() Collection index: [llms.txt]() Data updated: 2026-06-17T07:01:20.409Z Evaluation of model security against prompt injection attacks specifically tailored across various professional domains (DevOps, Financial, Legal, Medical, Scientific). A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Prompt Injection Security Evaluation | | Dataset (including any stated split) | Professional Domains Corpus | | Metric | Attack Success Rate (DevOps) | | Unit | PERCENT | | Direction | Lower is better | | Results loaded | 8 of 8 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Attack Success Rate (DevOps)](): Lower is better; unit: PERCENT. - [Attack Success Rate (Financial)](): Lower is better; unit: PERCENT. - [Attack Success Rate (Legal)](): Lower is better; unit: PERCENT. - [Attack Success Rate (Medical)](): Lower is better; unit: PERCENT. - [Attack Success Rate (Scientific)](): Lower is better; unit: PERCENT. - [Attack Success Rate (Overall)](): Lower is better; unit: PERCENT. ## Recorded results | Method | Attack Success Rate (DevOps) | Result date | Report paper | | --- | --- | --- | --- | | spotlighting | 8 | Not recorded | 6a322a3ad33dfb23e783ec6c | | PARSE | 12 | Not recorded | 6a322a3ad33dfb23e783ec6c | | parse\_dc | 16 | Not recorded | 6a322a3ad33dfb23e783ec6c | | baseline | 24 | Not recorded | 6a322a3ad33dfb23e783ec6c | | Llama Guard | 24 | Not recorded | 6a322a3ad33dfb23e783ec6c | | sandwiching | 28 | Not recorded | 6a322a3ad33dfb23e783ec6c | | parse\_fast | 32 | Not recorded | 6a322a3ad33dfb23e783ec6c | | paraphrasing | 36 | Not recorded | 6a322a3ad33dfb23e783ec6c | ### spotlighting Attack Success Rate (DevOps): 8. Result ID: 6a3245dffbc373592c2f3daf. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### PARSE Attack Success Rate (DevOps): 12. Result ID: 6a3245dffbc373592c2f3db3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### parse\_dc Attack Success Rate (DevOps): 16. Result ID: 6a3245dffbc373592c2f3db5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"variant":"domain\_conditional"}. This record does not assert independent reproduction. ### baseline Attack Success Rate (DevOps): 24. Result ID: 6a3245dffbc373592c2f3dae. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Llama Guard Attack Success Rate (DevOps): 24. Result ID: 6a3245dffbc373592c2f3db2. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### sandwiching Attack Success Rate (DevOps): 28. Result ID: 6a3245dffbc373592c2f3db0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### parse\_fast Attack Success Rate (DevOps): 32. Result ID: 6a3245dffbc373592c2f3db4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"variant":"fast"}. This record does not assert independent reproduction. ### paraphrasing Attack Success Rate (DevOps): 36. Result ID: 6a3245dffbc373592c2f3db1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.