Investigating how to improve AI Infrastructure for finding and resolving security issues within systems before deployed, with exact reproduction measurements
Inspired by Semgrep's recent tool, and a rise in understanding that going forward for many years, we will be requested to cater to Security measures a lot, as hacking and intrusion will become common occurrence
To build defensible AI techniques, and models tuned, and tools for those models, to detect minute CVEs internally before release -- as attackers have further capacity to attack and hurt the average everyday individual.