WebsiteHunt is broughtBrought to you by Redact Everything
SS
Small-Samples Poison in LLMs
Anthropic research on data-poisoning attacks in LLMs.
This page summarizes Anthropic's research on data-poisoning in large language models. The study shows that as few as 250 malicious documents can backdoor models from 600M to 13B parameters, challenging the idea that poisoning scales with data volume. It discusses backdoors, risks, and directions for mitigation and defense.
Comments
Loading comments...
Category
You might also like
HiAPI
One API for leading AI image, video, and text generation models. ✨ Premium
ProjectManagementTools
Optimize your Projects with ProjectManagementTools 🌱 Sponsored
Labrynth AI
AI regulatory intelligence platform 🌱 Sponsored
Anthropic: AI Safety & Policy
Responsible AI research and policy updates from Anthropic.
Jina Reader API
Read URLs and search web for better grounding LLMs.
Frontiers in Political Science
Open-access, peer-reviewed research in political science.
Buzz App
AI-powered Market Research
Introducing Claude Opus 5
Anthropic's model for long-running agents and coding.
Position: LLMs can't jump
A position paper on LLMs, abduction, and scientific invention.
Stantem
Property Search, Real Estate Data API & Data Aggregation Services
NH
New Hormone Builds Strong Bones
Researchers uncover a hormone strengthening bones in lactation.
Wiz Blog
Security research and vulnerability disclosures in AI.
PlasticList
Data on plastic chemicals in Bay Area foods
Korvus
Search SDK to unify RAG pipeline in a single database query
ChatPulse
Gain real insights on team communication by leveraging Slack data
RamenLegal
AI suite for legal documentation and research.
TimeCapsuleLLM
An LLM trained only on era-specific data to reduce modern bias.
Elicit
The AI research assistant
Juno Labs Blog
Thoughtful AI insights on privacy, data, and policy.
StratChat Premium
An AI-driven search engine for stocks, and investing
DataFuel.dev
Turn websites into LLM-ready data.
Keyword Insights
Save hours doing keyword research
GG