Publications
We've found a new class of vulnerabilities & threats facing LLMs and their deployment. Prepare for malware that can infect a language model!
I benchmark various different optimization algorithms on their fuzzing capabilities, to find that constraint solving with optimization might not have been the key to Angoras performance (popular fuzzing paper that year).
Exposing security problems of unsecured databases, predicted the wave of ransomware that took down the databases years later.
Projects
[
[