Jan Zahálka https://validator.w3.org/feed/docs/rss2.html zahalka.net: AI & security blog Backdoor attacks & defense @ CVPR ’23: How to build and burn Trojan horses Welcome! Can ChatGPT read who you are? A cyberattacker’s little helper: Jailbreaking LLM security Reality can be lying: Deepfakes and image manipulation @ CVPR ’23 What’s more powerful than one adversarial attack? Elves explain how to understand adversarial attacks Privacy attacks @ CVPR ’23: How to steal models and data AI security @ CVPR ’23: Honza’s highlights & conclusion Teaching Cheatsheet of AI security papers from CVPR ’22 Publications From “maybe” to “absolutely sure”: Certifiable security at CVPR ’23 How to see properly: Adversarial defense by data inspection AI security @ CVPR ’22: Non-classic adversarial attacks research Judging LLM security: How to make sure large language models are helping us? Subscribe AI security @ CVPR ’23 Subscription successful! AI security @ CVPR ’22: Model inversion attacks research Email confirmation Adversarial training: a security workout for AI models From one model to another: Transferable attacks research @ CVPR ’23 Better model architecture, better adversarial defense New adversarial attacks on computer vision from CVPR ’23 Which model architecture is the best in adversarial defense? Real-world AI security: Physical adversarial attacks research from CVPR ’23 From AI security to the real world: Physical adversarial attacks research @ CVPR ’22 The best AI security papers from CVPR ’23: Official highlights AI security @ CVPR ’22: Image manipulation & deepfake detection research AI security @ CVPR ’22: Backdoor/Trojan attacks research Are there guarantees in AI security? Certifiable defense research @ CVPR ’22 AI & security @ CVPR ’22: Classic adversarial attacks research How secure is computer vision? AI & security at CVPR ’22 How to intuitively understand adversarial attacks on AI models Sample Page Privacy policy