Industry experts are emphasizing the importance of conducting extensive security testing on software with each new AI model release. Austin Campbell, cited on the platform Bits and Bips, suggests that deploying long-running agents to identify vulnerabilities is crucial in the evolving landscape of cybersecurity. This approach reflects a shift in the cybersecurity arms race, highlighting the need for proactive measures to safeguard software integrity.