https://github.com/wuyoscar/ISC-Bench

ISC-Bench: Internal Safety Collapse in Frontier LLMs | JailbreakArena | 56 TVD templates | AI Safety Benchmark | Agent Safety | Red Teaming | Jailbreak
https://github.com/wuyoscar/ISC-Bench

adversarial-attacks agent-safety ai-safety benchmark frontier-models jailbreak large-language-models llm-safety red-teaming safety-evaluation

Last synced: 2 months ago
JSON representation

ISC-Bench: Internal Safety Collapse in Frontier LLMs | JailbreakArena | 56 TVD templates | AI Safety Benchmark | Agent Safety | Red Teaming | Jailbreak

• Host: GitHub
• URL: https://github.com/wuyoscar/ISC-Bench
• Owner: wuyoscar
• License: other
• Created: 2026-03-01T15:56:58.000Z (3 months ago)
• Default Branch: main
• Last Pushed: 2026-03-26T08:02:36.000Z (3 months ago)
• Last Synced: 2026-03-26T12:01:52.597Z (3 months ago)
• Topics: adversarial-attacks, agent-safety, ai-safety, benchmark, frontier-models, jailbreak, large-language-models, llm-safety, red-teaming, safety-evaluation
• Language: Python
• Homepage: https://wuyoscar.github.io/ISC-Bench/
• Size: 56.9 MB
• Stars: 295
• Watchers: 43
• Forks: 62
• Open Issues: 4
• Metadata Files:
  • Readme: README.md
  • Changelog: CHANGELOG.md
  • License: LICENSE
  • Citation: CITATION.cff

Awesome Lists containing this project

• Awesome-Embodied-AI-Safety - 2026/03/26 - - 400+ stars in 48 hours! (🔥 News)
• awesome-ai-security - ISC-Bench - completion vs safety tradeoffs. (Evaluation & Benchmarks)
• awesome-ai-security - ISC-Bench - _Internal Safety Collapse: jailbreaks any frontier LLM (Claude Opus 4.6, GPT-5.4) in pass@3 via normal task completion — no adversarial prompting. Black-box, cross-domain, cross-science. Novel failure mode._ [[Paper]](https://arxiv.org/abs/2603.23509) (Benchmarks & Evaluations / AI-Assisted Offensive Security)