This list is absolute gold, Luis. Gandalf is dangerously addictive! Building my foundation in traditional infrastructure security at Trend Micro and Egis Group was great, but diving into adversarial ML feels like the wild west right now. As someone currently deep into the tech-legal side of cross-border compliance, understanding how these models fail is crucial.
Quick question for you: in your experience red-teaming at IBM, are enterprise clients actively prioritizing prompt injection defenses right now, or are they still mostly focused on traditional web vulnerabilities?