In 1997, when IBM's supercomputer, Deep Blue, defeated world champion chess grandmaster Garry Kasparov, nobody concluded that chess computers should be banned. Instead, players studied them, analyzed their strategies, and learned from them. Three decades later, today's grandmasters play at a level unimaginable in the 1990s; not despite chess-playing supercomputers, but because of them.

Fast forward to the present, where organizations face a similar crossroads with advanced AI systems. For example, in June, the U.S. government temporarily banned Anthropic’s specialized cybersecurity model, Fable, sparking debate among industry experts and policymakers. The ban, triggered by national security concerns about the product’s capabilities, was lifted within a few weeks.

While concerns over increasingly sophisticated AI models are understandable, the history of chess suggests that technology bans are the wrong response. Banning the use of AI in cybersecurity will only make proper use of AI for defensive implementations more difficult, while malicious actors can benefit from its capabilities. Instead of treating advanced AI as a potential threat, it should be viewed as a training opportunity for the next generation of cybersecurity professionals, researchers, and defenders. Mastery comes from embracing technology and learning how to work with it, not from avoiding it completely.

The Rise of Specialized AI

Conversations around AI have focused on large, general-purpose models that can answer questions, generate content, and assist with everyday tasks. But now, the industry is moving in a different direction. Rather than building a single model that excels at many things, developers are creating specialized AI systems that perform exceptionally well within specific domains. By narrowing their scope, these systems become more useful, more precise, and better able to assist professionals with highly technical work.

Cybersecurity represents a clear example of this trend. Organizations face an expanding attack surface, while adversaries continue to automate their operations and shorten the time between discovering a vulnerability and exploiting it. As a result, security teams are often overwhelmed by an ever-growing volume of vulnerabilities, misconfigurations, and software flaws. Because even small improvements in cybersecurity efficiency can yield significant real-world benefits, specialized AI could become a powerful tool for helping organizations defend against threats.

A Chess Lesson for Cybersecurity

When chess programs first surpassed the world's strongest human players, they didn't make grandmasters obsolete. Instead, they fundamentally changed how players studied the game. Professionals began reviewing computer-generated variations, examining unconventional moves, and discovering strategic concepts that human analysis alone had overlooked. The result was that human players did not become obsolete; they became stronger by using computer analysis to discover new ideas and deepen their understanding of the game.

This history demonstrates the importance of recognizing that defenders need opportunities to understand the latest technology if they’re going to be equipped to defend against attackers who use it.

Any technology capable of identifying security vulnerabilities and gaps in an organization’s defenses naturally raises concerns that malicious actors could also misuse it to exploit those weaknesses. Those concerns should not be dismissed. Responsible safeguards, rigorous testing, and thoughtful oversight all play important roles as increasingly capable AI systems are integrated into cybersecurity operations and the tools used to design, build, and maintain digital infrastructure.

Just as chess players became stronger when they learned to work alongside powerful computer systems, security professionals can become more effective by using AI to analyze complex environments, uncover connections between vulnerabilities, and accelerate investigations into potential threats.

The Best Defense is a Better Defender 

If a security engineer is reviewing thousands of lines of application code, a specialized AI assistant can rapidly identify potential weaknesses, including improper input validation, privilege escalation paths, misconfigured access controls, and vulnerabilities introduced through software dependencies. Beyond flagging issues, the assistant can explain the underlying risk, prioritize findings by potential impact, and recommend remediation approaches. Instead of spending valuable time triaging large volumes of routine findings, security engineers can focus on complex architectural risks, threat modeling, and defending against novel attack techniques.

Over time, this engineer will become better at their job, not because the AI performed the work independently, but because it accelerated the learning process. This distinction can get lost in discussions about ways to stymie, or in some cases ban, certain AI technologies. The debate surrounding Anthropic’s Fable illustrates this challenge: restricting access may reduce certain risks, but it may also limit defenders’ ability to understand and responsibly apply the same capabilities to defend critical systems. When efforts are made to prevent AI misuse, it’s important to recognize that using it responsibly can strengthen an organization's overall security.

As AI advances, the next generation of security professionals will grow stronger by learning alongside these systems. Like the chess grandmasters who embraced powerful engines as training partners, security engineers who master these tools will be prepared for the challenges ahead. The strongest defenders will be the ones who study how technologies work, understand their strengths and limitations, and incorporate them into better defensive strategies.

That is what happened in chess. Instead of fearing computer analysis, elite players embraced it and became students of the technology. After that, human expertise advanced because humans gained access to a better challenger and turned it into a training partner.

The Future of Cybersecurity Depends on More Than AI

Increasingly specialized AI systems will not eliminate the need for experienced professionals. On the contrary, they will make human judgment, creativity, and strategic thinking even more valuable by removing routine tasks and accelerating technical analysis.

As such, conversations about AI governance should extend beyond the immediate risks posed by individual models. They should also consider the long-term consequences of limiting the cybersecurity community's ability to learn and adapt.

Building more capable AI is only part of the challenge. Equally important is equipping professionals with the knowledge to use these technologies effectively. If we want tomorrow's cybersecurity teams to defend against highly sophisticated AI-powered threats, they must have opportunities to study, be challenged by, and work alongside the technologies shaping the future of cyber defense.

Just as chess engines helped create a new generation of grandmasters, specialized AI can help create a new generation of cybersecurity experts, enhanced by the intelligent systems they use every day.