{"slug": "fuck-your-ai-safeguards", "title": "Fuck Your AI Safeguards", "summary": "A senior product engineer writing on ounapuu.ee reported that Claude Code blocked his attempt to fix a security vulnerability in his own product, requiring him to apply for Anthropic's Cyber Verification Program exemption, which he said involves waiting multiple business days, forfeiting data retention exemptions, and submitting a government ID or passport with no guarantee of approval. He said he instead used EU-hosted open weights models, including GLM 5.3, to confirm the issue, fix it, and verify the resolution. The post argues that criminals can use abliterated open weights models without safeguards, so Anthropic's restrictions leave defenders worse off.", "body_md": "# [Fuck your AI safeguards](https://ounapuu.ee/posts/2026/11/19/unsafeguards/)\n\n[While refrigerators were getting hacked in Estonia,](https://news.err.ee/1610183296/hackers-raised-freezer-temperature-at-selver-supermarket-to-50-c)\nI half-seriously took a project I’m working on and asked Claude Code if there are any low-hanging fruit regarding\nsecurity that we could improve. It found one issue that was easy to fix, simple oversight with a one line fix, so I just\nwent along with it.\n\nThen it got stopped because of bullshit “AI safeguards”. Oh no, how dare ***I***, the senior product engineer with\n***my*** product that I provably have control over, try to *** fix*** and prevent security issues and ensure compliance\nwith local laws and regulations?\n\n“Oh but you can just apply to get an exemption from them here *bla bla fucking blah*”\n\nHave you **READ**[the requirements?](https://support.claude.com/articles/14604842-cyber-verification-program)\n\nFirst of all, you will have to wait for multiple business days to get a response.\n\nYou forfeit any data retention exemptions, so hope you don’t have any IP to protect!\n\nOh, and oh so ***please*** give us your fucking ID or passport! We won’t leak it later, promise!\n\nAnd you may just get fucked anyway and not get approval. **Approval to appropriately protect your own service!**\n\nThis wasn’t the first time I ran into this issue. As maddening as it sounds, I *try* to keep up with the security\nlandscape and to apply appropriate mitigations because leaking sensitive information is ***a big fucking no-no*** and I\ncan’t let that happen.\n\nIf there’s a popular exploit out there that applies for our technical stack, I need to know if the public\nproof-of-concept snippets can be adapted and used maliciously against my services. But no, Claude will tell you to fuck\noff and use an older model that *still* will try to stop you from applying adequate protection.\n\nYou know who don’t need to ask for approval? Criminals. They just go and use the models that don’t have safeguards in place. Abliterated open weights models don’t care. Bad actors reach their goals with or without your “safe” models.\n\nYou can use a knife to cut up some food, or seriously hurt someone. Yet I don’t get asked to put in an application with my government-provided ID to buy one. Why is this tool so special?\n\nPredictably, the only way to ***actually*** make our own service more secure ended up involving the use of EU-hosted\nopen weights models, such as GLM 5.3, that actually complied with our intentions, proved that the issue is there, and\nthen helped fix and verify that the issue is now resolved, for good.\n\nCongratulations, Anthropic, you’ve made the world an unsafer place.\n\nOr maybe that was your plan all along to get regulators to react and build that moat for you right before your IPO to fend off competitors?\n\nEither way, I’m just incredibly frustrated. Big AI labs use the same tools without restrictions to make the security\nlandscape a fucking mess, and then they go and prohibit others from using the same tools to protect themselves from the\nthreats they just created. Fucking asshats.[1](#fn:1)\n\n1. and that’s my swear word quota hit for the year. [↩︎](#fnref:1)\n\nSubscribe to new posts via [the RSS feed.](https://ounapuu.ee/index.xml)\n\nNot sure what RSS is, or how to get started? [Check this guide!](https://aboutfeeds.com/)\n\nYou can reach me via [e-mail](mailto:ihavesomethoughtsonyourblog@ounapuu.ee) or [LinkedIn.](https://www.linkedin.com/in/%F0%9F%94%A5-herman-%C3%B5unapuu-600516152/)\n\nIf you liked this post, consider sharing it!", "url": "https://wpnews.pro/news/fuck-your-ai-safeguards", "canonical_source": "https://ounapuu.ee/posts/2026/11/19/unsafeguards/", "published_at": "2026-11-19 16:30:18+00:00", "updated_at": "2026-11-19 16:43:42.272578+00:00", "lang": "en", "topics": ["ai-safety", "ai-policy", "large-language-models", "ai-tools"], "entities": ["Anthropic", "Claude Code", "GLM 5.3", "Herman Õunapuu", "Cyber Verification Program"], "also_reported_by": [], "takeaway": null, "recommendation": null, "alternates": {"html": "https://wpnews.pro/news/fuck-your-ai-safeguards", "markdown": "https://wpnews.pro/news/fuck-your-ai-safeguards.md", "text": "https://wpnews.pro/news/fuck-your-ai-safeguards.txt", "jsonld": "https://wpnews.pro/news/fuck-your-ai-safeguards.jsonld"}}