Claude Code blocks a security fix, so a developer switches to GLM
What happened
On November 20, Hacker News front-paged a report that a senior product engineer using Claude Code to fix a security flaw in his own product was stopped by "AI safeguards" and could not complete a one-line fix. He switched to GLM 5.3, an open-weight model hosted in the EU, and verified and fixed the issue. He criticized the exemption process, which takes several working days and requires waiving data-retention exemptions and submitting ID. He also noted attackers can use open models with no safeguards.
Written by AI from the coverage · updated 22 hours ago
Coverage
Follow the reports to see the story from different sides.
- Hacker News front pageFuck Your AI Safeguards
一位资深产品工程师在用 Claude Code 修复自家产品的安全漏洞时,因“AI safeguards”拦截而无法完成一行代码的修复。他最终改用欧盟托管的开源权重模型 GLM 5.3,成功验证并修复了该安全问题。他批评申请豁免需等待多个工作日、放弃数据保留豁免并提交身份证件,而攻击者可直接使用无 safeguards 的开源模型。
Heat over time
Heat now 5·Comparable peak 9(Nov 20 03:00)·Comparable change over 24 hours –
The trend only compares accounts observed without gaps, so its range may be smaller than the current heat. Hover or tap the chart for each hour; the left and right arrow keys step through it.