The debate is not limited to the United States.
The United Nations has been discussing autonomous weapons and artificial intelligence for years, with governments debating how international law should apply to increasingly sophisticated technologies.
A 2025 UN document highlighted several risks associated with military AI, including excessive reliance on AI for critical military decisions, vulnerabilities to manipulation and cyber interference, and the possibility that poorly integrated systems could produce unintended consequences.
The UN has also emphasized principles such as human responsibility, accountability, weapons reviews and compliance with international humanitarian law when dealing with autonomous weapons systems.
In September 2025, Guterres again called for lethal autonomous weapons to be prohibited when they operate without human control and stressed that decisions involving nuclear weapons must remain with humans rather than machines.
These discussions reveal how far the issue has moved beyond science fiction.
Governments are now debating questions that Cameron dramatized on screen decades ago:
Who is responsible when an automated system makes a catastrophic decision?
How much autonomy should a machine have?
How can humans intervene?
How can countries prevent an AI competition from becoming destabilizing?
And what international rules should apply when different countries develop different standards?
AI Does Not Have to Become Conscious to Be Dangerous
One misconception surrounding The Terminator comparison is that AI would need to become self-aware before it could pose a serious threat.
That is not necessarily the case.
A system does not need emotions, consciousness or hatred to create harmful consequences.
A sufficiently powerful automated system could cause problems simply because it follows poorly defined objectives, receives inaccurate information or operates under inappropriate assumptions.
This is why modern AI safety discussions often focus on concepts such as alignment, reliability, robustness, oversight and human control.
The concern is essentially about ensuring that the system's behavior remains compatible with human intentions and legal requirements.
The challenge becomes harder as systems become more complex.
If humans cannot fully understand why an AI system produced a particular output, it becomes harder to determine whether that system should be trusted in a high-stakes environment.
The problem becomes even more complicated when multiple AI systems interact with one another.
Instead of one fictional Skynet controlling everything, imagine a world containing thousands of specialized automated systems developed by different organizations, operating under different objectives and connected to increasingly complex infrastructure.
That scenario is much less cinematic—but potentially more difficult to govern.
The AI Arms Race Could Be Driven by Fear
Cameron's comparison to nuclear competition also points toward a psychological problem.
Countries may not develop powerful military AI because they want to create an uncontrollable future.
They may develop it because they are afraid someone else will do so first.
This is the classic security dilemma.
One country's attempt to increase its security can make another country feel less secure. The second country responds by developing its own capabilities. The first country sees that response as confirmation that it needs even more advanced systems.
The cycle continues.
Cameron described this dynamic directly when he suggested that if one side does not develop military AI, another side eventually will, creating pressure for escalation.
This is why international cooperation matters.
Technology itself does not automatically create an arms race. Competition between states can.
Agreements, transparency measures, testing standards, communication channels and clear boundaries could potentially reduce some of those risks.