DigitalToday reporter Chi-gyu Hwang - British offensive security startup CodeWall has raised $4 million in funding. CodeWall provides AI agents to banks and other companies that attempt to attack them directly like external hackers.

SiliconANGLE reported on Oct. 14 local time that CodeWall's agents start attacks without account credentials or internal prior information. Multiple agents with different roles operate at the same time and compile what they find into a single investigation map.

If one agent finds a weakness in an app, other agents use it as a foothold to penetrate connected systems and data.

Customers can verify attack paths step by step, with evidence. The agents attack again each time a new system is introduced or a security patch is applied. They operate only within the scope permitted by the customer. CodeWall's strategy is to replace penetration tests that companies typically commission once or twice a year with this service.

CodeWall stresses that defenders also need the automation attackers already use. Earlier this summer, cloud security company Sysdig disclosed the first ransomware attack case carried out from start to finish by an AI agent.

CodeWall CEO Paul Price (폴 프라이스) said, "Businesses do not need to wait for a real attacker to find out whether their defenses are working properly."