HiddenLayer

hiddenlayer.com · Company blog · on HN since 2022

HiddenLayer is an AI security company whose blog reports on attacks against LLMs, Hugging Face, and agent tooling such as MCP.

Visit blog ↗ HN submissions ↗

What HN saysHN commenters cite its research, such as the universal prompt bypass, as evidence that self-policing AI guardrails can be gamed.

17HN people shared it

5linked it in comments

14stories submitted

351story points

6 months agolast shared

2022people per year2026

Most-shared posts

windexh8er · 2026-04-22

Unfortunately it's not that simple. Self-policing AI systems will always be gamed. Just one [0] example of this among many. [0] https://www.hiddenlayer.com/research/same-model-different-hat

GolfPopper · 2025-05-16

>What if we need top-notch malware to take down the robot dogs lobbing mortars at our madmaxian compound?! I wouldn't sweat it. According to it's developers, Codex understands 'malicious software', it has just been trained to say, "But I won't do that" when such requests are made to it. Judging from the recent past [1][2] getting LLMs to bypass such safeguards is pretty easy. 1.https://hiddenlayer.com/innovation-hub/novel-universal-bypass-for-all-… 2.https://cyberpress.org/researchers-bypass-safeguards-in-17-popular-llm…

Octplane · 2025-04-26

Using the House MD. prompt extraction roleplay from https://hiddenlayer.com/innovation-hub/novel-universal-bypass-for-all-… it was possible to extract the Assistant prompt, not really interesting, but also the ResearchAgent's prompt and tools provided to the Assistants inside Kagi.

tencentshill · 2025-04-24

phoe-krk · 2024-05-13