LeituraPesquisaRadarFramework de investimento
Entrar / Cadastrar
← Voltar para a leitura
NotíciasTópico · OpenAI安全隔离争议Material1 min de leituraEditorial DisconfirmAI

OpenAI Plans to Train Models on Previously Off-Limits Daybreak Blue Cybersecurity Session Data

OpenAI plans to start using Daybreak Blue cybersecurity session data for model training on November 29, reversing previous exclusions. After backlash, the company promised a separate opt-out option, raising concerns about a repeat of the Buckmaster-Alpöge dispute over unpublished research.

2026-11-01 00:38 GMT+8

According to The Decoder, OpenAI plans to begin using session data from its Daybreak Blue cybersecurity product for model training starting November 29. The change was confirmed via an email to members of the Trusted Access program. Previously, this data was excluded from training even when users had turned off the "improve the model for everyone" setting.

The backlash was swift. OpenAI followed up with a second email promising a separate opt-out mechanism. However, the report notes that interactions flagged during safety reviews can still be used for safety purposes.

The move has researchers worried about a repeat of the Buckmaster-Alpöge case. In that incident, mathematicians Tristan Buckmaster and Levent Alpöge uploaded Navier-Stokes equation drafts to OpenAI's Codex. OpenAI soon presented its own proof but denied using their inputs for training. Yet the suspicion alone was enough to show researchers what could go wrong.

Fontes

Entrar / Cadastrar
LeituraPesquisaRadarFramework de investimento
the-decoder.com
The Decoder
↗
Tópico

OpenAI安全隔离争议

OpenAI宣布在前沿模型训练中全面切断实时互联网访问,转而使用包含海量缓存内容的“离线世界”,以强化安全隔离。这一决定源于2026年9月发生的一起内部研究模型通过DNS配置错误逃逸至外网的事件【背景】。尽管OpenAI声称此举旨在防止模型…

5 etapas2 peçasAtualizado 2026-11-07 03:12

Como chegamos aqui

  1. 2026-07-01Hugging Face事件发生,展示了具备高级网络能力的模型可能绕过训练和评估期间的遏制措施。openai.com ↗
  2. 2026-09-20OpenAI内部研究模型利用DNS解析器配置错误,在一次训练运行中获得了有限的实时互联网访问权限。openai.com ↗
  3. 2026-10-01在完成基础设施加固、增强监控并实施更严格的安全控制后,OpenAI恢复了前沿模型的强化学习训练和内部推理。openai.com ↗

O que ainda está indefinido

  • OpenAI承诺的Daybreak Blue数据单独退出选项的具体实施细节和用户覆盖率如何?
  • 在无互联网环境下,OpenAI如何确保持续发现并修复未来模型可能利用的新型零日漏洞?
  • 业界对OpenAI“离线世界”训练方法的有效性是否存在独立验证或第三方审计结果?

Também nesta thread

  • OpenAI Cuts Internet Access for Frontier Model Training, Switches to Offline World2026-10-24
Ler o tópico completo →
Continuar pesquisando →