At CPX APAC 2025, Dan Karpati, VP of AI at Check Point, shared insights on the current state of AI in cybersecurity and where it's headed. He outlined the journey from AI-assisted security to fully autonomous systems and discussed the obstacles that organizations must overcome to make this transition.
Artificial intelligence is transforming cybersecurity, offering organizations a path from manual defenses to fully autonomous threat mitigation. As cyber threats grow in sophistication, AI is no longer just an assistive tool but is gradually evolving into an independent decision-maker in security operations. However, challenges remain – particularly in balancing automation with human oversight.
At CPX APAC 2025, Dan Karpati, VP of AI at Check Point, shared insights on the current state of AI in cybersecurity and where it's headed. He outlined the journey from AI-assisted security to fully autonomous systems and discussed the obstacles that organizations must overcome to make this transition.
The road to fully autonomous security
Cyber threats are evolving at an unprecedented rate, with AI-powered phishing attacks, deepfake fraud, and adaptive malware challenging traditional defenses. Check Point is working to develop AI-driven security systems capable of proactively countering these emerging threats.
Karpati outlined the key phases of AI-driven cybersecurity evolution:
• Manual Security: Human-driven threat detection and response.
• AI-Assisted Security: AI automates repetitive tasks and provides insights.
• AI-Augmented Security: AI enhances human decision-making with predictive analytics.
• Supervised-AI Autonomy: AI operates independently under human supervision.
• Full Security Autonomy: AI detects, mitigates, and resolves threats without human intervention.
The ultimate goal is to eliminate human bottlenecks, allowing AI to autonomously manage network security, However, Karpati still sees this as a distant future: “In the long run I think humans always want to have control I don’t think there will be a point that machines will have total control, since it contradicts the human nature” he emphasized.
The current state: AI-assisted security
According to Karpati, the cybersecurity industry is currently in the "AI-assisted security" stage. AI is reactive, providing insights and recommendations while humans remain in control. However, we are moving toward the "AI-augmented" phase, where AI agents will take on more tasks autonomously, with humans shifting into an oversight role.
“The movement will be to an augmented phase, where work will be partially done by AI autonomously,” Karpati explained. “More tasks will be handled by AI agents, and as we evolve, security operations will become increasingly autonomous. However, I believe a fully autonomous system is a futuristic goal that may never be fully achieved. Nonetheless, we should set it a goal to strive to.”
The balancing act: AI vs. human expertise
While AI can automate many cybersecurity functions, human expertise remains critical. Karpati envisions AI functioning similarly to a workplace hierarchy, where AI starts as a junior-level assistant, learning and growing in responsibility over time. “Over time, humans will focus more on orchestration—guiding AI systems rather than handling individual security tasks,” he noted.
However, this progression also raises ethical concerns. AI decision-making in cybersecurity involves risks, especially when dealing with human-related security incidents. “There are open ethical issues that need to be resolved, as these systems are making decisions that impact people,” Karpati cautioned.
AI’s game-changer: Generative AI
Karpati pointed out that generative AI (Gen AI) has marked a turning point in cybersecurity. Unlike traditional AI, which relied on domain-specific proprietary data, Gen AI leverages vast amounts of human-generated data, enabling it to provide more nuanced and adaptive solutions.
“Our previous security engines were built on proprietary data,” he explained. “Now, Gen AI allows organizations to use human language to create endless use cases. It brings AI closer to how humans think, even writing code on the fly.”
While this opens up unprecedented opportunities, it also introduces risks—such as securing dynamically generated code. “The ability of AI to generate backend code on the fly is both tempting and risky,” Karpati warned. “We need to implement AI-driven security measures to protect against AI-generated threats.”
Conclusion: The future is AI-driven, but human-guided
The transition to autonomous cybersecurity is not just a technological shift—it’s a paradigm shift in how organizations approach digital defense. While AI is set to take on greater responsibility, human oversight will remain crucial in guiding its evolution.
“The future is fascinating,” Karpati reflected. “AI innovation allows underdogs to redefine the game, just as DeepSeek breakthroughs disrupted OpenAI with its creativity.”
As AI continues to evolve, businesses must embrace both its potential and its challenges. Those who adapt to this new era of AI-driven security will be best positioned to stay ahead of increasingly sophisticated cyber threats.
Product Adopted:Other