LLMs corrupt your documents when you delegate (arxiv.org) | HN Screenshots

YHacker Newsscreenshots

Screenshot of arxiv.org: LLMs corrupt your documents when you delegate

First screen, captured 5 months ago.

Looks similar

Screenshot of arxiv.org: LLMorphism: When humans come to see themselves as language models

LLMorphism: When humans come to see themselves as language models (arxiv.org)

60 points by okey 5 months ago | 39 comments

Screenshot of arxiv.org: Alignment pretraining: AI discourse creates self-fulfilling (mis)alignment

Alignment pretraining: AI discourse creates self-fulfilling (mis)alignment (arxiv.org)

67 points by anigbrowl 5 months ago | 25 comments

Screenshot of arxiv.org: A sleep-like consolidation mechanism for LLMs

A sleep-like consolidation mechanism for LLMs (arxiv.org)

178 points by juxtapose 5 months ago | 126 comments

Screenshot of arxiv.org: Teuken-7B-Base and Teuken-7B-Instruct: Towards European LLMs

Teuken-7B-Base and Teuken-7B-Instruct: Towards European LLMs (arxiv.org)

235 points by doener 1 year ago | 88 comments

Screenshot of arxiv.org: Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs

Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs (arxiv.org)

168 points by martythemaniak 1 year ago | 43 comments

Screenshot of arxiv.org: Unsupervised Elicitation of Language Models

Unsupervised Elicitation of Language Models (arxiv.org)

128 points by kordlessagain 1 year ago | 18 comments

Related stories

Screenshot of news.mit.edu: LLMs use a surprisingly simple mechanism to retrieve some stored knowledge

LLMs use a surprisingly simple mechanism to retrieve some stored knowledge (news.mit.edu)

336 points by CharlesW 2 years ago | 121 comments

MIT News article on research showing LLMs often retrieve stored facts via simple linear functions, and a technique to probe what a model knows about subjects.

large-language-models interpretability knowledge-retrieval mit

Screenshot of simonwillison.net: 2025: The Year in LLMs

2025: The Year in LLMs (simonwillison.net)

823 points by simonw 9 months ago | 451 comments

Simon Willison's annual blog post reviewing major LLM trends of 2025, including reasoning models, coding agents, Chinese open weight models, and Gemini.

llms ai-agents claude-code coding-agents

Screenshot of bleepingcomputer.com: New Linux udisks flaw lets attackers get root on major Linux distros

New Linux udisks flaw lets attackers get root on major Linux distros (bleepingcomputer.com)

336 points by smig0 1 year ago | 227 comments

BleepingComputer news article about a udisks vulnerability that lets attackers gain root privileges on major Linux distributions.

linux udisks privilege-escalation vulnerability

Screenshot of interconnects.ai: GLM-5.2 is a step change for open agents

GLM-5.2 is a step change for open agents (interconnects.ai)

344 points by vantareed 4 months ago | 197 comments

Nathan Lambert's Interconnects post arguing that the GLM-5.2 model marks a capability threshold for open-weight agentic AI.

glm-5-2 open-weight-models ai-agents zhipu-ai

Screenshot of simonlermen.substack.com: Large-Scale Online Deanonymization with LLMs

Large-Scale Online Deanonymization with LLMs (simonlermen.substack.com)

327 points by DalasNoin 8 months ago | 233 comments

Substack post by Simon Lermen summarizing research showing LLM agents can deanonymize users from anonymous posts across Hacker News, Reddit, LinkedIn and interviews at scale.

deanonymization privacy llm-agents osint

Screenshot of alexhwoods.com: Do your own writing

Do your own writing (alexhwoods.com)

690 points by karimf 7 months ago | 222 comments

Blog post by Alex Woods arguing against using LLMs to write documents and essays, since writing is a way to think and build trust.

llm writing ai-generated-content technical-writing