🕳 LLM Vulnerabilities - emschwartz

Less-relevant results

Phishing for Lobsters: How We Tricked OpenClaw into Spilling Secrets

💉Prompt Injection Blog

varonis.com··Hacker News

Pliny the Liberator 🐉󠅫󠄼󠄿󠅆󠄵󠄐󠅀󠄼󠄹󠄾󠅉󠅭 (@elder_plinius)

🎭Claude

xcancel.com··Hacker News

Measuring Embedding Drift: Why Hybrid Search Saves Stale Models.

🔗Hybrid Search

pub.towardsai.net

jianzhichun/permafrost: Freeze Claude Code's prompt prefix so DeepSeek's automatic cache always hits — alignment proxy + coalescing + keepalive, installable as a CC plugin. Measured 64% cheaper on real Claude Code traffic.

🔌Claude Plugins Code

github.com··Hacker News

New ChatGPT Lockdown Mode Limits Tools That Could Enable Data Exfiltration

💉Prompt Injection

thehackernews.com·

Toward Secure LLM Agents: Threat Surfaces, Attacks, Defenses, and Evaluation

💉Prompt Injection Academic

arxiv.org·

Report: GKE Inference Gateway delivers up to 92% faster AI responses

🤖AI Blog

cloud.google.com··Hacker News

You Can Catch Sleeper Agents by Teaching Another Model to Imitate Them

💉Prompt Injection

lesswrong.com·

OpenAI rolls out a Lockdown Mode for extra protection against prompt injection attacks

🛡️AI Security News

engadget.com·

Apple's AI Can Now Change Your Passwords. What Could Possibly Go Wrong?

🛡️AI Security Blog

kylereddoch.me··Hacker News

Anthropic says these topics are too dangerous to let its Fable 5 model talk about

🎭Claude News

arstechnica.com·

Making Local LLM Go Brrr

🤖AI

seanpedersen.github.io·

Show HN: Built an open-source local firewall for AI coding agents

💻Coding Agents Discussion

news.ycombinator.com··Hacker News

How to Build an Agentic RAG with RubyLLM and Rails

🏗️Search Infrastructure Blog

panasiti.me··Hacker News

Context-Fractured Decomposition Attacks on Tool-Using LLM Agents: Exploiting Artifact Provenance Gaps

💉Prompt Injection Academic

arxiv.org·

Polymarket Annotation Injection

🛡️AI Security

sam.elborai.me··Hacker News

How we fight GPU scarcity without compromise

🏗️LLM Infrastructure Blog

equixly.com··Hacker News

How Edgar Allan Poe found bugs in Turso

🔱LibSQL Blog

turso.tech··Hacker News

Meet Hades: The malware that lies to AI security agents

OpenAI expands ChatGPT Lockdown Mode for sensitive data

Phishing for Lobsters: How We Tricked OpenClaw into Spilling Secrets

Pliny the Liberator 🐉󠅫󠄼󠄿󠅆󠄵󠄐󠅀󠄼󠄹󠄾󠅉󠅭 (@elder_plinius)

Measuring Embedding Drift: Why Hybrid Search Saves Stale Models.

jianzhichun/permafrost: Freeze Claude Code's prompt prefix so DeepSeek's automatic cache always hits — alignment proxy + coalescing + keepalive, installable as a CC plugin. Measured 64% cheaper on real Claude Code traffic.

New ChatGPT Lockdown Mode Limits Tools That Could Enable Data Exfiltration

Toward Secure LLM Agents: Threat Surfaces, Attacks, Defenses, and Evaluation

Report: GKE Inference Gateway delivers up to 92% faster AI responses

You Can Catch Sleeper Agents by Teaching Another Model to Imitate Them

OpenAI rolls out a Lockdown Mode for extra protection against prompt injection attacks

Apple's AI Can Now Change Your Passwords. What Could Possibly Go Wrong?

Anthropic says these topics are too dangerous to let its Fable 5 model talk about

Making Local LLM Go Brrr

Show HN: Built an open-source local firewall for AI coding agents

How to Build an Agentic RAG with RubyLLM and Rails

Context-Fractured Decomposition Attacks on Tool-Using LLM Agents: Exploiting Artifact Provenance Gaps

Polymarket Annotation Injection

How we fight GPU scarcity without compromise

How Edgar Allan Poe found bugs in Turso