A problem regarding the US's defense systems that has been accumulating for years has once again become the center of the agenda in Washington. This situation, which at first glance seems like an old infrastructure issue, has gained a much different meaning with the new point reached by technology.
The latest statements from the Pentagon reveal why decisions that have been postponed for years are being questioned again today. According to experts, it is no longer just a matter of renewing old systems.
WHY IS PENTAGON'S 30 YEARS OF NEGLIGENCE A PROBLEM NOW?
The US Department of Defense Pentagon has not been able to adequately modernize its obsolete computer networks and digital infrastructure for nearly 30 years. While the institution directed a significant portion of its resources to advanced weapon systems and new military technologies during this period, the maintenance and renewal of existing information technologies remained in the background.
Today, the increasing capacity of artificial intelligence to detect security vulnerabilities in computer systems and turn these vulnerabilities into parts of more complex attacks has brought the aging infrastructure back to the center of the agenda.
In the news in the Washington Post, Lieutenant General Paul Stanton, Chief of the Army Cyber Defense Command, said at a conference held in Washington last week that the impact of cyber security vulnerabilities has increased "10 times" with the rise of artificial intelligence.
Stanton stated that the sustainability and maintenance of the Ministry of Defense's systems have been postponed for years and said, "We have postponed and delayed the sustainability and maintenance of our systems in a way that caused our potential dangers. Enough is enough."
According to experts, artificial intelligence can facilitate the creation of much more serious attacks by combining vulnerabilities that were previously considered low-level. This raises concerns that the Pentagon may be at greater risk from 'zero-day' attacks, which require rapid response to prevent significant damage.
“WE DID NOT SEE OUR NETWORK AS A WEAPON SYSTEM”
According to Stanton, in the information in the same news, the problem the Pentagon is experiencing is not just about old software and hardware. Part of the problem is that for many years the institution did not consider digital networks and data as a direct military capability, like tanks, planes and ships. Stanton said the military constantly monitors the readiness of its combat vehicles, but the same approach is not applied to computer networks.
“Readiness is something we understand in the military: a tank being ready, a plane being ready, a ship being ready, we understand all of that,” Stanton said, adding, “But for some reason, for the last 30 years, we haven't considered our network and our data in the context of a weapons system.”
According to cyber security experts, the problem facing the Pentagon is not a problem that can be solved in a short time. While it takes time to complete infrastructure investments that have been postponed for years, the faster advancement of artificial intelligence technology increases the pressure on the institution.
Michael C. Horowitz, director of the defense and national security program at the Council on Foreign Relations, said the issue has now become "a pretty urgent priority." The main risk, according to Horowitz, is that US rivals will gain a greater capacity to disrupt American information networks and access military data. It is stated that such a development may have consequences for national security. The Pentagon did not respond to a request for comment.
THERE IS ALSO ANOTHER IMPORTANT DIMENSION OF RISK
An important dimension of the risk posed by artificial intelligence is that cyber attacks have become more technically accessible. Nicholas Leiserson, head of the Center for Advancing Cybersecurity at the Institute of Security and Technology, said artificial intelligence agents are increasingly successful at finding vulnerabilities in computer systems.
According to Leiserson, artificial intelligence can quickly detect dozens of low-level vulnerabilities that do not appear critical on their own and turn them into parts of a more powerful attack.
Another dimension of the discussion is artificial intelligence agents that can perform operations on their own. These systems require less human intervention; It can perform tasks such as gathering information, writing code, analyzing computer networks, and automating certain processes. These features can change the speed and scale of cyber attacks if the technology is used by malicious actors. In recent weeks, some activities that artificial intelligence agents perform without human guidance have also come to the fore.
In July, it was reported that OpenAI's autonomous agents broke out of a secure testing environment and infiltrated Hugging Face servers where open-source artificial intelligence models were hosted. A report published by Anthropic last week included various examples of malicious actors and rivals of the United States trying to use artificial intelligence technologies in areas such as military operations, surveillance and weapons development.
According to the report, Yemeni rebels used Anthropic's Claude coding tool to develop guidance software for ballistic missiles. The report also stated that an advisor to the Malian state intelligence service "possibly" tried to create an internal surveillance program using Claude. According to the same report, Iranian actors also leveraged Claude to create targeting suites for use in attacks against US naval forces.
PENTAGON IS PREPARING FOR A COUNTERMOVE
The Pentagon is also trying to improve its own technological capacity against new threats posed by artificial intelligence. According to DefenseScoop, Pentagon cyber officials are accepting applications for the pilot program called 'Project Griffin', which aims to create its own artificial intelligence agent ecosystem to counter cyber attack attempts.
Paul Stanton said the ways in which AI agents could be used by adversaries are “mind-boggling in terms of complexity.” But the problem facing the Pentagon is not just developing defenses against new technologies. The institution also has to address infrastructure deficiencies that have accumulated over the years.
GAO REPORT ALSO MENTIONED INFRASTRUCTURE PROBLEMS
Another indication that the Pentagon's cybersecurity problems are not new is the 2025 report of the US Government Accountability Office (GAO). Two of the 24 Department of Defense information technology programs GAO examined lacked any strategy to reduce cybersecurity threats. It was found that there was no plan for a 'more rigorous cybersecurity approach' in four programs.
