Terra Security has launched an automatic firewall rule generation feature, dubbed Prevention, to block attacks that AI agents have verified as workable.

SiliconANGLE reported on Aug. 3 that the feature helps cover the gap between confirming a vulnerability and reflecting code fixes in production environments.

Prevention works by having an AI agent first confirm whether a specific vulnerability can be exploited, then checking whether controls already in operation at a customer work along that attack path. If controls are insufficient, it generates web application firewall (WAF) or general firewall rules and then revalidates whether the exploit is blocked. A security team decides whether to apply the changes in production, and sensitive changes require human approval.

Terra Security said patches can take weeks to be applied because of release cycles and change-management approval procedures. It said disclosed vulnerabilities can be weaponised within minutes, underscoring the need for blocking rules to fill the gap.

Existing penetration test reports stopped at recommending mitigation measures such as firewall tuning, network segmentation and configuration changes. The company said they could not specify what rules to write or whether those rules actually block the exploits described in the reports.

Gaal Malachi (갈 말라치), co-founder and chief technology officer of Terra Security, said finding exploitable risks alone is not enough. He said the hardest part is the period after proving them and before making actual fixes, and that Prevention helps defence organisations secure response time in between.