Cybersecurity Execs: AI Threats Outpacing Defensive Measures

Executive Summary

During the Asia New Vision Forum on September 29, 2026, prominent cybersecurity leaders warned that the pace of artificial intelligence (AI) development is enabling a new class of cyber threats that corporate defenses are not prepared to handle. Gaurav Keerthi, CEO of StrongKeep Cybersecurity, stated that the threat landscape is evolving beyond traditional data breaches to include systemic risks like supply chain attacks and physical harm from AI-enabled technologies such as smart cars and robots. The consensus was that the emergence of autonomous hacking agents and the use of AI to accelerate attacks are creating a significant capabilities gap between attackers and defenders.

Threat Overview

The core concern raised by industry experts is that AI is not just another tool but a force multiplier for threat actors. Key points from the discussion include:

• Autonomous Hacking Agents: AI systems are being developed that can operate with increasing independence, capable of identifying vulnerabilities, crafting exploits, and executing attacks with minimal human intervention. This dramatically increases the speed and scale at which attacks can occur.
• Expansion of Threat Surface: As AI is integrated into more physical systems (e.g., IoT devices, autonomous vehicles, industrial robots), the potential impact of a cyberattack shifts from data loss to kinetic, real-world harm.
• Supply Chain as a Primary Vector: Attackers are targeting the AI development lifecycle itself, seeking to poison training data or compromise AI models before they are even deployed, creating widespread downstream risk.
• Defensive Lag: The development of defensive AI technologies is perceived to be lagging behind the weaponization of offensive AI, leaving organizations vulnerable.

These warnings are substantiated by recent events. A September 30 report on the cyberattack against Spain's rail network detailed how attackers used AI to exfiltrate 500 GB of data in mere hours. Additionally, AI company Anthropic recently reported on its efforts to disrupt malicious actors using its Claude AI models for cyber operations between late 2025 and August 2026, confirming that the abuse of commercial AI for malicious purposes is already happening.

Impact Assessment

The strategic implications of this trend are profound:

• Compressed Response Times: Security Operations Centers (SOCs) that rely on human-speed analysis and response will be overwhelmed by machine-speed attacks. The detection-to-response window is shrinking to minutes or seconds.
• Novel Attack Vectors: Traditional security tools may not be effective against attacks that manipulate AI model behavior or exploit vulnerabilities in AI-specific software stacks.
• Increased Systemic Risk: A single compromised AI model or poisoned dataset used by thousands of organizations could lead to a cascading, global failure, far exceeding the impact of traditional software vulnerabilities.
• Erosion of Trust: As AI systems become more autonomous, ensuring they are secure and acting as intended becomes a paramount challenge, with failures potentially leading to a public backlash against the technology.

Detection & Response

Defending against AI-powered threats requires a paradigm shift towards AI-powered defense.

• AI for Anomaly Detection: Use machine learning and AI to baseline normal behavior across networks, endpoints, and user activity to spot the subtle deviations that may indicate an AI-driven attack.
• Automated Response: Implement Security Orchestration, Automation, and Response (SOAR) platforms to enable machine-speed responses to detected threats, such as automatically isolating a compromised host.
• AI Model Security: Develop new techniques for monitoring the integrity and behavior of deployed AI models to detect signs of tampering or unexpected outputs.

Mitigation Recommendations

• Secure AI Development Lifecycle (SAIL): Organizations developing or deploying AI must adopt a security-first mindset, embedding security controls throughout the entire AI lifecycle, from data acquisition to model retirement.
• Assume Breach of AI Systems: Design systems with the assumption that AI components could be compromised. Implement zero-trust architectures and strict segmentation to limit the blast radius.
• Invest in Defensive AI Research: The cybersecurity community and governments must increase investment in research and development of AI-based defensive technologies to keep pace with offensive capabilities.
• Collaboration and Information Sharing: Rapid sharing of threat intelligence related to AI-powered attacks is crucial for the collective defense of the ecosystem.