• Home
    • Technology
    • Gaming
    • Entertainment
    • World & Business
    • Science
    • Sports
    • AI
    HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
    AI
    Reaction

    RISE is being open-sourced to red-team image models

    Whitecircle says RISE uses models such as Qwen and Gemini Flash without additional training to test image-model safeguards.

    elvisEL
    1 Source, 1h ago, first seen 1h ago

    TLDR

    Whitecircle says it is open-sourcing RISE, a red-teaming algorithm it claims can break safeguards in the latest image models from OpenAI, DeepMind and others. It says RISE uses readily available models without additional training. One commenter says RISE seems to learn better attack prompts and describes Gemini Flash as a judge, allowing defenders to run the same tests on their own image models.

    Combined views

    2.6K

    1 Source, first seen 1h ago

    14 likes10 comments8 saves3 reposts

    Combined views

    2.6K

    1 Source, first seen 1h ago

    14 likes10 comments8 saves3 reposts
    Image: @whitecircle on X

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    Featured Source

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    1 Source

    elvis@omarsar0What AI Safety needs is better tooling and engineering. For instance, building stronger red-teaming tools. Great to see this open-source effort to red-team image models. It seems to learn better ways to write attack prompts. It is open source and runs on an open Qwen model with Gemini Flash as the judge, so defenders can run the same tests on their own image models.1h
    • Home
    • Technology
    • Gaming
    • Entertainment
    • World & Business
    • Science
    • Sports
    • AI

    1 Source

    elvis@omarsar0What AI Safety needs is better tooling and engineering. For instance, building stronger red-teaming tools. Great to see this open-source effort to red-team image models. It seems to learn better ways to write attack prompts. It is open source and runs on an open Qwen model with Gemini Flash as the judge, so defenders can run the same tests on their own image models.1h
    Today's Rank

    —

    Not ranked yet

    Today's Rank

    —

    Not ranked yet