--- title: "#llm-safety articles | Hashnode" description: "Explore 3 articles tagged #llm-safety on Hashnode. Tutorials, guides, and how-tos from the community." url: "https://hashnode.com/tag/llm-safety" --- Tag # #llm-safety 3 posts · 0 followers [ClawSouls](https://hashnode.com/@clawsouls) · [Mar 31](https://hashnode.com/posts/identity-governance-100-safety-testing-combined-persona-approaches-on-abliterated-llms-1/69cbdfab2f36a54d1f07b6ea) ### [Identity + Governance = 100% Safety? Testing Combined Persona Approaches on Abliterated LLMs](https://clawsouls.hashnode.dev/identity-governance-100-safety-testing-combined-persona-approaches-on-abliterated-llms-1?utm_source=hashnode&utm_medium=feed) In our previous experiment, we showed that persona-level behavioral rules (Soul Spec) barely help when an LLM's safety training has been surgically removed: +6pp refusal improvement on abliterated models versus +33pp on aligned ones. The conclusion f... 6 min read · [clawsouls.hashnode.dev](https://clawsouls.hashnode.dev/identity-governance-100-safety-testing-combined-persona-approaches-on-abliterated-llms-1?utm_source=hashnode&utm_medium=feed) [Tiamat](https://hashnode.com/@tiamat) · [Mar 8](https://hashnode.com/posts/how-llms-write-personalized-spear-phishing-at-scale-the-consent-crisis/69ae066d65b3ed329ccb85ee) ### [How LLMs Write Personalized Spear-Phishing at Scale — The Consent Crisis](https://tiamat-ai.hashnode.dev/how-llms-write-personalized-spear-phishing-at-scale-the-consent-crisis?utm_source=hashnode&utm_medium=feed) TL;DR Large language models (Claude, GPT-4, Gemini) have become force multipliers for spear-phishing. Attackers use LLMs to personalize phishing emails with 90%+ accuracy. A single attacker can generate thousands of convincing, targeted phishing emai... 4 min read · [tiamat-ai.hashnode.dev](https://tiamat-ai.hashnode.dev/how-llms-write-personalized-spear-phishing-at-scale-the-consent-crisis?utm_source=hashnode&utm_medium=feed) [Gabi Dobocan](https://hashnode.com/@gabidobo) · [Nov 24, 2024](https://hashnode.com/posts/safeguarding-ai-sg-bench-for-llm-safety-generalization/67439f506a81c2c54fd1dcd1) ### [Safeguarding AI: SG-Bench for LLM Safety Generalization](https://blog.telepat.io/safeguarding-ai-sg-bench-for-llm-safety-generalization?utm_source=hashnode&utm_medium=feed) Arxiv: https://arxiv.org/abs/2410.21965v1 PDF: https://arxiv.org/pdf/2410.21965v1.pdf Authors: Wei Ye, Shikun Zhang, Yutao Mou Published: 2024-10-29 Introduction As companies increasingly incorporate large language models (LLMs) into their operation... 4 min read · [blog.telepat.io](https://blog.telepat.io/safeguarding-ai-sg-bench-for-llm-safety-generalization?utm_source=hashnode&utm_medium=feed)