home.social

#adversarial-attacks — Public Fediverse posts

Live and recent posts from across the Fediverse tagged #adversarial-attacks, aggregated by home.social.

fetched live
  1. Comprehensive guide to LLM security threats and defenses. Learn how attackers exploit AI models and practical strategies to protect against adversarial attacks. hackernoon.com/adversarial-att #adversarialattacks

  2. North Carolina State University: New Attack Can Make AI ‘See’ Whatever You Want. “Researchers have demonstrated a new way of attacking artificial intelligence computer vision systems, allowing them to control what the AI ‘sees.’ The research shows that the new technique, called RisingAttacK, is effective at manipulating all of the most widely used AI computer vision systems.”

    https://rbfirehose.com/2025/07/08/north-carolina-state-university-new-attack-can-make-ai-see-whatever-you-want/

  3. Adversarial attacks pose serious threats to ML systems. 🤖 Lumenova AI's blog explores cutting-edge detection & defense strategies to safeguard AI. 🛡️ From adversarial training to ensemble methods, learn how to build robust models. 🏗️ #AdversarialAttacks #MachineLearning #AIRobustness

    🔗 s.42l.fr/nJ0rqqxA

    Bckp.:

    lumenova.ai/blog/adversarial-a

  4. 📣 Presentation at Glücksspielsymposium #SympGS24Forschungsstelle Glücksspiel (Universität Hohenheim)

    🎲 Adding to notions of #nudging and #DarkPatterns I spoke about other forms of Human-Technology Relations through #AdversarialAttacks, #Jailbreaking, and #SpecificationGaming

    👾 Technical objects always afford more than what is intended by providers in their marketing logic. The spaces of possibility that lie beyond the logic of exploitation in the nature of the object indicate what will happen sooner or later in the application of technologies anyway- and the better you know your way around, the easier it is to help shape, counteract, be creative, make informed decisions and act in a self-determined way.

    What's more, dealing with these things is simply a lot of fun. Thank you very much for the invitation and the positive feedback! The slides will soon be available on the University of Hohenheim website.

    #genAI #objectrecognition #philosophy #philosophyoftechnology

  5. @CommieGIR
    Somersaulting, cardboard box, and tree disguise, as simple #adversarialAttacks in human detection #AI #algos