Oct 10 • 7 months ago
AI models can acquire backdoors from surprisingly few malicious documents
Anthropic study suggests "poison" training attacks don't scale with model size. ...
{"_":"https://arstechnica.com/ai/2025/10/ai-models-can-acquire-backdoors-from-surprisingly-few-malicious-documents/","$":{"isPermaLink":"true"}}1 min read
Read More