安全研究员炮轰 Anthropic 扩版 CVP:防御侧工作 95% 被排除在外
npinto · x · 2026-10-07
- Anthropic announced it expanded its Cyber Verification Program: OSS maintainers and individual researchers with a track record of reported vulns can now qualify for the Defense Access tier.
- But a security researcher publicly criticized the program, saying roughly 95% of defensive cybersecurity work is excluded from CVP and that the product has become nearly unusable even with it, especially now that access focuses on vulnerability researchers.
- The researcher says they've shifted to GLM — a sign CVP's gating may push real practitioners toward rival models.
「公司和人」频道最新
- 前 OpenAI 高管吐槽:每次刷 LinkedIn 都看到 OpenAI 广告 — Miles_Brundage · 2026-10-07
- Transformer 论文共同作者 Llion Jones:走红带来的烦恼不少 — YesThisIsLion · 2026-10-07
- IMDEA 研究:Claude 网页版向 Meta、TikTok 等第三方回传用户事件 — ZoeyPanthera · 2026-10-07
- 观察:营销团队仍沉迷 Canva,而非直接用 ChatGPT 出图 — gaganghotra_ · 2026-10-07
- Basis Fellowship 开放申请:10 分钟申请、无条件 5 万美元 — ahandvanish · 2026-10-07
- Liquid AI×Zetic 端侧 AI 工作坊:无 Wi-Fi 也能跑的应用 — helloiamleonie · 2026-10-07