As cybercriminals are increasingly using artificial intelligence (AI) to find weaknesses in company systems and launch attacks faster, cybersecurity companies are simultaneously developing AI-powered tools that can test how easily hackers could break into an organisation’s systems.
One of these cybersecurity companies, Armadin, has secured $255.5 million in Series B funding to expand its technology, which uses AI agents to simulate cyberattacks against companies before real attackers can exploit their security weaknesses.
Announced in early October, the funding round was co-led by Andreessen Horowitz and Accel, with participation from Bain Capital Ventures, Redpoint, Google Ventures, Kleiner Perkins, Menlo Ventures, In-Q-Tel, 8VC, and Ballistic Ventures.
The investment brings Armadin’s total funding to $445 million and pushes its valuation above $2.5 billion, just seven months after the company publicly launched in March 2026.
How Armadin Uses AI to Hack Companies
Founded by Kevin Mandia, the cybersecurity entrepreneur behind Mandiant, alongside Travis Lanham, Evan Peña and David Slater, Armadin is developing a security platform that uses autonomous AI agents to identify weaknesses in company networks. Mandiant was also previously founded by Mandia, which Google acquired for $5.4 billion in 2022.
Armadin’s technology works by deploying thousands of specialised AI agents that behave like hackers, attempting to break into systems, gain access to sensitive areas, and identify possible routes attackers could use.
This approach goes beyond traditional penetration testing, where security professionals are hired to simulate cyberattacks and identify vulnerabilities at a particular point in time.
Armadin’s AI agents can continuously examine company systems, identify multiple weaknesses, and test whether those weaknesses can be combined to gain deeper access. The findings will then help security teams understand which vulnerabilities could lead to a successful attack and where they need to strengthen their protection.
Armadin’s AI Agents Have Already Conducted Millions of Offensive Actions in One Test
The company has already demonstrated how its technology works through a cybersecurity exercise conducted with TENEX.ai in August 2026.
During the three-day exercise, Armadin deployed 26,000 AI agents against the systems of an unnamed global institution with its permission.
According to the companies, the agents carried out 17 million offensive actions, launched 1,300 attacks, and identified 38 verified attack paths alongside 238 security findings.
Meanwhile, TENEX.ai’s security platform analysed more than 101,000 alerts generated during the exercise, demonstrating how AI could be used to investigate attacks happening at such a large scale.
Although the results were reported by the companies involved, the exercise provides an example of how Armadin’s technology is being used to identify security weaknesses in live environments.
Where Armadin’s $255 Million Funding Goes Next
Armadin plans to use the new investment to expand its AI-powered security platform, strengthen research and training, as well as support efforts to bring its technology to more customers.
The company says it is already conducting AI-driven security testing for Fortune 500 companies and government organisations.
The funding also comes as investors increase their backing for cybersecurity startups developing technology to address AI-driven threats, especially as Armadin’s approach reflects a growing concern within the cybersecurity industry that AI could allow attackers to discover and exploit vulnerabilities faster than security teams can respond.
By testing systems continuously and showing companies how their weaknesses could be exploited, Armadin is betting that organisations will increasingly need to identify potential attack routes before cybercriminals discover them.
As such, the company’s next challenge will be expanding that technology across more organisations while demonstrating how effectively its AI-powered testing helps them prevent real-world cyberattacks.
