Stopping a cyberattack while walking your dog - defensive AI security CEO says it's not ruff to do
Corma CEO tells The Reg it's building 'One ring to rule them all, for the defenders to have this power'
Alon Pluda, CEO of AI security startup Corma, emphasizes the importance of closing the defensive gap in cybersecurity. The startup's goal is to develop models that are better at defensive security than their offensive counterparts, which can now find and fix bugs and orchestrate tools across complex workflows. Pluda uses the analogy of walking his dog to illustrate a successful defensive security action - a security executive who received a notification on his smartwatch about a live attack while out for a walk.
The executive approved the agent's action, and the model blocked the malware, mitigating the intrusion within ten minutes. Corma, founded last year, received $60 million in seed funding, and the company is working with Fortune 100 companies, training models to achieve "superintelligence for defensive cybersecurity." The company tested four frontier models against each other in attacker and defender roles, finding that the models were successful at implanting backdoors 85% of the time, but only detected 19% of attacks.
Pluda believes that defensive security involves tasks that don't involve scanning code for vulnerabilities and misconfigurations, and these tasks are less reliable for models trained on data that is neither prose nor source.
Written by urgent.news from The Register Science's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
This story
This is one outlet's version. Read the fullest account.
