Tag: prompt injection attacks
-
Hacker News: Garak, LLM Vulnerability Scanner
Source URL: https://github.com/NVIDIA/garak Source: Hacker News Title: Garak, LLM Vulnerability Scanner Feedly Summary: Comments AI Summary and Description: Yes **Summary:** The text describes “garak,” a command-line vulnerability scanner specifically designed for large language models (LLMs). This tool aims to uncover various weaknesses in LLMs, such as hallucination, prompt injection attacks, and data leakage. Its development…
-
Simon Willison’s Weblog: Quoting Question for Department for Science, Innovation and Technology
Source URL: https://simonwillison.net/2024/Nov/1/prompt-injection/#atom-everything Source: Simon Willison’s Weblog Title: Quoting Question for Department for Science, Innovation and Technology Feedly Summary: Lord Clement-Jones: To ask His Majesty’s Government what assessment they have made of the cybersecurity risks posed by prompt injection attacks to the processing by generative artificial intelligence of material provided from outside government, and whether…
-
The Register: Google reportedly developing an AI agent that can control your browser
Source URL: https://www.theregister.com/2024/10/28/google_ai_web_agent/ Source: The Register Title: Google reportedly developing an AI agent that can control your browser Feedly Summary: Project Jarvis will apparently conduct research, purchase products, and even book a flight on your behalf Google is reportedly looking to sidestep the complexity of AI-driven automation by letting its multimodal large language models (LLMs)…
-
The Register: Anthropic’s latest Claude model can interact with computers – what could go wrong?
Source URL: https://www.theregister.com/2024/10/24/anthropic_claude_model_can_use_computers/ Source: The Register Title: Anthropic’s latest Claude model can interact with computers – what could go wrong? Feedly Summary: For starters, it could launch a prompt injection attack on itself… The latest version of AI startup Anthropic’s Claude 3.5 Sonnet model can use computers – and the developer makes it sound like…
-
Simon Willison’s Weblog: Quoting Model Card Addendum: Claude 3.5 Haiku and Upgraded Sonnet
Source URL: https://simonwillison.net/2024/Oct/23/model-card/#atom-everything Source: Simon Willison’s Weblog Title: Quoting Model Card Addendum: Claude 3.5 Haiku and Upgraded Sonnet Feedly Summary: We enhanced the ability of the upgraded Claude 3.5 Sonnet and Claude 3.5 Haiku to recognize and resist prompt injection attempts. Prompt injection is an attack where a malicious user feeds instructions to a model…
-
Simon Willison’s Weblog: This prompt can make an AI chatbot identify and extract personal details from your chats
Source URL: https://simonwillison.net/2024/Oct/22/imprompter/#atom-everything Source: Simon Willison’s Weblog Title: This prompt can make an AI chatbot identify and extract personal details from your chats Feedly Summary: This prompt can make an AI chatbot identify and extract personal details from your chats Matt Burgess in Wired magazine writes about a new prompt injection / Markdown exfiltration variant…
-
Hacker News: Hacker plants false memories in ChatGPT to steal user data in perpetuity
Source URL: https://arstechnica.com/security/2024/09/false-memories-planted-in-chatgpt-give-hacker-persistent-exfiltration-channel/ Source: Hacker News Title: Hacker plants false memories in ChatGPT to steal user data in perpetuity Feedly Summary: Comments AI Summary and Description: Yes Summary: The text discusses a vulnerability discovered in ChatGPT that allowed for malicious manipulation of its long-term memory feature through prompt injection. While OpenAI has released a partial…