Experimental News Clipping Site

Tag: prompts

  • Wired: Psychological Tricks Can Get AI to Break the Rules

    —

    by

    Kurt Seifried
    in

    Source URL: https://arstechnica.com/science/2025/09/these-psychological-tricks-can-get-llms-to-respond-to-forbidden-prompts/ Source: Wired Title: Psychological Tricks Can Get AI to Break the Rules Feedly Summary: Researchers convinced large language model chatbots to comply with “forbidden” requests using a variety of conversational tactics. AI Summary and Description: Yes Summary: The text discusses researchers’ exploration of conversational tactics used to manipulate large language model (LLM)…

  • The Register: Shell to pay: Crims invade your PC with CastleRAT malware, now in C and Python

    —

    by

    Kurt Seifried
    in

    Source URL: https://www.theregister.com/2025/09/05/clickfix_castlerat_malware/ Source: The Register Title: Shell to pay: Crims invade your PC with CastleRAT malware, now in C and Python Feedly Summary: Pro tip, don’t install PowerShell commands without approval A team of data thieves has doubled down by developing its CastleRAT malware in both Python and C variants. Both versions spread by…

  • Schneier on Security: GPT-4o-mini Falls for Psychological Manipulation

    —

    by

    Kurt Seifried
    in

    Source URL: https://www.schneier.com/blog/archives/2025/09/gpt-4o-mini-falls-for-psychological-manipulation.html Source: Schneier on Security Title: GPT-4o-mini Falls for Psychological Manipulation Feedly Summary: Interesting experiment: To design their experiment, the University of Pennsylvania researchers tested 2024’s GPT-4o-mini model on two requests that it should ideally refuse: calling the user a jerk and giving directions for how to synthesize lidocaine. The researchers created experimental…

  • Docker: You are Doing MCP Wrong: 3 Big Misconceptions

    —

    by

    Kurt Seifried
    in

    Source URL: https://www.docker.com/blog/mcp-misconceptions-tools-agents-not-api/ Source: Docker Title: You are Doing MCP Wrong: 3 Big Misconceptions Feedly Summary: MCP is not an API. Tools are not agents. MCP is more than tools. Here’s what this means in practice. Most developers misread the Model Context Protocol because they map it onto familiar API mental models. That mistake breaks…

  • Schneier on Security: Indirect Prompt Injection Attacks Against LLM Assistants

    —

    by

    Kurt Seifried
    in

    Source URL: https://www.schneier.com/blog/archives/2025/09/indirect-prompt-injection-attacks-against-llm-assistants.html Source: Schneier on Security Title: Indirect Prompt Injection Attacks Against LLM Assistants Feedly Summary: Really good research on practical attacks against LLM agents. “Invitation Is All You Need! Promptware Attacks Against LLM-Powered Assistants in Production Are Practical and Dangerous” Abstract: The growing integration of LLMs into applications has introduced new security risks,…

  • New York Times – Artificial Intelligence : How to Rethink A.I.

    —

    by

    Kurt Seifried
    in

    Source URL: https://www.nytimes.com/2025/09/03/opinion/ai-gpt5-rethinking.html Source: New York Times – Artificial Intelligence Title: How to Rethink A.I. Feedly Summary: Building bigger A.I. isn’t leading to better A.I. AI Summary and Description: Yes Summary: The assertion that creating larger AI systems does not necessarily enhance their efficacy highlights a critical issue within the field of artificial intelligence. This…

  • Slashdot: One Long Sentence is All It Takes To Make LLMs Misbehave

    —

    by

    Kurt Seifried
    in

    Source URL: https://slashdot.org/story/25/08/27/1756253/one-long-sentence-is-all-it-takes-to-make-llms-misbehave?utm_source=rss1.0mainlinkanon&utm_medium=feed Source: Slashdot Title: One Long Sentence is All It Takes To Make LLMs Misbehave Feedly Summary: AI Summary and Description: Yes Summary: The text discusses a significant security research finding from Palo Alto Networks’ Unit 42 regarding vulnerabilities in large language models (LLMs). The researchers explored methods that allow users to bypass…

  • Slashdot: Google Improves Gemini AI Image Editing With ‘Nano Banana’ Model

    —

    by

    Kurt Seifried
    in

    Source URL: https://slashdot.org/story/25/08/26/215246/google-improves-gemini-ai-image-editing-with-nano-banana-model?utm_source=rss1.0mainlinkanon&utm_medium=feed Source: Slashdot Title: Google Improves Gemini AI Image Editing With ‘Nano Banana’ Model Feedly Summary: AI Summary and Description: Yes Summary: Google DeepMind has launched the “nano banana” model (Gemini 2.5 Flash Image), which excels in AI image editing by offering improved consistency in edits. This advancement enhances the practical use cases…

  • Simon Willison’s Weblog: Piloting Claude for Chrome

    —

    by

    Kurt Seifried
    in

    Source URL: https://simonwillison.net/2025/Aug/26/piloting-claude-for-chrome/#atom-everything Source: Simon Willison’s Weblog Title: Piloting Claude for Chrome Feedly Summary: Piloting Claude for Chrome Two days ago I said: I strongly expect that the entire concept of an agentic browser extension is fatally flawed and cannot be built safely. Today Anthropic announced their own take on this pattern, implemented as an…