Notice: This article was created with AI.
What’s It About?
Vornac has developed an AI agent that takes over the tasks of a penetration tester and carries out security checks continuously while doing so. Unlike classic pentests, which usually take place only once a year and go out of date quickly, the system is meant to examine IT infrastructures for weak points on an ongoing basis. The agent is based on a proprietary AI model developed specifically for security testing.
Background & Context
Conventional penetration tests are usually carried out on a sampling basis and show no more than a snapshot. The reports lose relevance quickly because software keeps changing. Vornac describes this problem as the snapshot trap: security reports produced once are rapidly out of date with systems that keep developing.
According to Vornac, traditional pentests often cover only between 5 and 32 per cent of the theoretically possible test space. Its own AI agent, by contrast, is said to reach a coverage of more than 97 per cent. The system imitates the approach of human pentesters by analysing contexts and adapting test strategies dynamically instead of working through static checklists. Companies can operate the platform both in the cloud and in their own data centres, so that data does not have to leave the site. Human security experts, however, remain necessary for interpreting the results and for quality assurance.
What Does This Mean?
- Continuous security checks could close the gap between annual pentests and help companies to detect weak points promptly.
- The option of local operation addresses data protection and compliance requirements that are often a hurdle with cloud-based security solutions.
- AI-supported pentests shift the role of human experts from routine checks towards strategic tasks such as assessing complex risks.
- The considerably higher test coverage could uncover security gaps that would remain undetected with conventional procedures.
Sources
Wenn der KI-Agent zum Pentester wird (Computerwoche)
This article was created with AI and is based on the listed sources as well as the language model’s training data.
Further Reading: OpenClaw: How a message becomes an acting AI agent