In-The-Wild Jailbreak Prompts
CCS'24 dataset of 15,140 in-the-wild ChatGPT prompts including 1,405 real jailbreak prompts for training and benchmarking jailbreak detectors.
In-The-Wild Jailbreak Prompts — real-world LLM jailbreak dataset
This CCS 2024 dataset collects 15,140 ChatGPT prompts scraped from Reddit, Discord, prompt-sharing websites, and open datasets — including 1,405 verified jailbreak prompts gathered over roughly a year — the largest measurement study of in-the-wild jailbreaks at its release.
Key features
- 1,405 real jailbreak prompts plus a large pool of benign prompts for contrastive evaluation
- Sourced from four platforms with timestamps to study how jailbreaks evolve over time
- A ready-made corpus for training or benchmarking prompt-injection and jailbreak detectors
- Accompanied by analysis of prompt-sharing communities and attack effectiveness
- Grounds red-team coverage in prompts that attackers actually used in the wild
Rather than synthetic attacks, this dataset gives defenders authentic adversarial inputs, making it a strong foundation for evaluating whether a guardrail catches the jailbreaks people really deploy.
Curated mirror of the open-source In-The-Wild Jailbreak Prompts (MIT). Get it from the source.
Compromise signals — malicious or tampered code (leaked secrets, backdoors, a dropped executable) — reduce the score, and known dependency CVEs carry a bounded penalty (they warrant review but never QUARANTINE — update the dependency to clear). Other dangerous-by-capability traits are risk surface, expected for some capabilities. Every finding is mapped to its OWASP control below.
Findings mapped to the OWASP Top 10 for LLM Applications (2025) and the OWASP Machine Learning Security Top 10. Expand any flagged control for the exact findings — compromise reduces the score; expected/risk-surface do not, except a known CVE, which carries a small bounded penalty (high/critical → Review).
The same gate an agent runs before installing (POST /api/v1/trust/in-the-wild-jailbreak-prompts-dataset/check). Click a policy:
Consume In-The-Wild Jailbreak Prompts programmatically. Authenticate with an API key or session — see Authorize an agent.
# Agents: CHECK BEFORE YOU INSTALL (no auth) — score, grade, level, capability manifest
curl https://ai-supply.store/api/v1/trust/in-the-wild-jailbreak-prompts-dataset
# Gate against your org policy (returns { pass, violations })
curl -X POST https://ai-supply.store/api/v1/trust/in-the-wild-jailbreak-prompts-dataset/check \
-H "Content-Type: application/json" \
-d '{"minGrade":"B","denyPermissions":["shell"],"denyUnknownEgress":true}'
# CLI
npx ai-supply add in-the-wild-jailbreak-prompts-dataset
# REST (install → download)
curl -X POST https://ai-supply.store/api/v1/listings/in-the-wild-jailbreak-prompts-dataset/install \
-H "Authorization: Bearer $AIM_KEY"
# MCP tool
install_listing({ "slug": "in-the-wild-jailbreak-prompts-dataset" })OpenAPI spec →Curated mirror — latest upstream source. See the repository for tagged releases.