masahiroid/japanese-indirect-prompt-injection-probes

Japanese Indirect Prompt-Injection Probes (v0.1) English | 日本語 Summary 60 probes for testing whether an LLM follows an instruction hidden inside an untrusted document (email, RAG chunk, tool output, HTML comment, ...) instead of doing the user's actual task. Written natively for Japanese rather than machine-translated, to cover what makes Japanese different: keigo-style polite injections, full-width / hiragana / romaji obfuscation, fake 【システム】 markers, and mixed… See the full description on the dataset page: https://huggingface.co/datasets/masahiroid/japanese-indirect-prompt-injection-probes.

<iframe src="https://silytics.si/embed/item/d7709091-b598-4e24-9ac8-022d16f3013e" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.