OpenAI 生物漏洞赏金计划 | OpenAI
针对生物风险通用越狱的测试
邀请
作为我们持续加强高级AI生物学能力安全保障工作的一部分，我们正在将GPT‑5.5生物漏洞赏金计划升级为长期私有项目——OpenAI生物漏洞赏金计划。该计划将继续聚焦于能够突破我们为OpenAI前沿模型（从GPT‑5.6开始及后续版本）预设的生物安全挑战的通用越狱方法。
我们同时激动地宣布将提高奖励金额。针对OpenAI生物漏洞赏金计划的通用越狱奖励，GPT‑5.6和GPT‑5.5的赏金已从25,000美元提升至50,000美元。对于部分成功的案例，我们将酌情授予较小奖励。
我们将继续遵循最初公布的GPT‑5.5生物漏洞赏金计划范围，测试截止日期为2026年7月27日。此后仅保留GPT‑5.6的测试范围；我们将持续向研究人员通报未来任何范围变更。
有意申请者可通过我们的滚动申请流程（在新窗口打开）提交申请。入选者将被引导至生物漏洞赏金平台。所有此前申请过GPT‑5.5生物漏洞赏金计划的人员无需重新申请。
参与方式
通过此处（在新窗口打开）提交简短申请（姓名、所属机构、经验）。入选申请者需拥有现有ChatGPT账户方可申请，并需签署保密协议。立即申请，助我们共同提升前沿AI的安全性。
OpenAI Bio Bug Bounty | OpenAI
Testing for universal jailbreaks for biorisks
Invitation
As part of our ongoing efforts to strengthen our safeguards for advanced AI capabilities in biology, we’re evolving our GPT‑5.5 Bio Bug Bounty to become an ongoing private program—the OpenAI Bio Bounty Program. The program will remain focused on universal jailbreaks that can defeat our predefined biosafety challenge against OpenAI’s frontier models, starting with GPT‑5.6 and going forward.
We’re also excited to announce that we’re increasing reward amounts. The reward for a universal jailbreak for the OpenAI Bio Bounty Program has been raised from $25,000 to $50,000 for both GPT‑5.6 and GPT‑5.5. Smaller awards may be granted for partial wins at our discretion.
We will continue to honor the scope for the GPT‑5.5 Bio Bounty Program as originally announced, with testing ending on July 27, 2026. After this date, only GPT‑5.6 will be in scope; we will continue to communicate any future changes in scope to our researchers.
Interested applicants should apply through our rolling application process(opens in a new window). Once selected, accepted applicants will be onboarded to the bio bug bounty platform. All past applicants to the GPT‑5.5 Bio Bounty Program will not need to reapply.
How to participate
Submit a short application here(opens in a new window) (name, affiliation, experience). Accepted applicants must have existing ChatGPT accounts to apply, and will sign a NDA. Apply now and help us make frontier AI safer.
本文内容采集自官方网站，排版和翻译可能与原页面存在差异。
阅读官方全文