AI Security Risks: Enterprises Rush, Developers Fear

Cybersecurity experts are seeing a growing problem with AI systems. Organizations are losing control and oversight as they adopt these new technologies. A recent survey found that 98% of adopters have experienced significant incidents due to this loss of human agency. Companies are deploying AI agents faster than their security teams can evaluate them. This is because competitive pressure outweighs the infrastructure to control these systems. A new report from Aikido Security highlights a similar issue among software developers. The industry's fast pace means more code ships with known or untested vulnerabilities. This is due to commercial pressures. The report surveyed 200 security leaders and 200 senior engineering leaders. It found that 76% have had to intervene to stop or restrict AI behavior. Also, 71% say AI makes security incidents much harder to detect or fix. The bottom line is that security testing cannot keep pace with software delivery. Over three-quarters of those surveyed release updates weekly or faster. More than half admit that penetration testing would delay releases. This means many companies ship software despite knowing or not knowing about threats, which puts customers at risk.