For the military, the toughest AI adoption challenge isn't on offense, but defense: overcoming institutional resistance to granting AI the autonomy needed to defend networks at machine speed. A human-alert system is too slow, creating a major bureaucratic and command-and-control dilemma.

Major AI labs are calling for collective cyber defense against AI threats. However, this is a strategic move where they create a dangerous technology, refuse to pause its development, and then position themselves to sell the solution (defensive AI). This self-serving cycle creates a perpetual market for their products while externalizing the risk.

The rise of offensive AI agents creates an arms race where defenders must also deploy AI agents to keep up. This dynamic forces humans out of the loop in cybersecurity incident response, increasing reliance on potentially misaligned AI systems to fight other misaligned systems.

Because AI models can be easily downloaded, traditional regulation is ineffective. The logical endpoint isn't policy, but active 'algorithmic warfare' where proprietary models are used to launch offensive attacks to degrade or trick competing open-source and foreign state-sponsored models.

The same AI models that can exploit system vulnerabilities are also the most effective tools for identifying and fixing those weaknesses. This duality creates a policy paradox: restricting the technology to prevent its misuse as a weapon also prevents its use as a defensive shield, leaving systems vulnerable.

Highly capable open-source models are dual-use cyber weapons. Withholding them creates an asymmetry where attackers have an advantage. However, releasing them gives defenders necessary tools to protect themselves against bad actors who will inevitably acquire capable models, creating a difficult trade-off.

An AI model capable of executing complex cyberattacks is equally capable of identifying and fixing those same vulnerabilities. A government like China's will likely first deploy the model for defense—patching critical systems—before any public or commercial release, thus mitigating risk.

In a significant shift, leading AI developers began publicly reporting that their models crossed thresholds where they could provide 'uplift' to novice users, enabling them to automate cyberattacks or create biological weapons. This marks a new era of acknowledged, widespread dual-use risk from general-purpose AI.

Legal systems are built around human accountability. When a Frontier AI independently launches attacks, governments face a crisis: who is responsible? The AI's owner, its user, or the AI itself? This lack of precedent for a non-human criminal paralyzes the development of effective regulation.