It’s Not AI, It’s People: Why This Sounds Like “Guns Don’t Kill People”

From gun rights to AI governance? And how a slogan built to block reform may be back in disguise.

Every time I bring up AI’s role in warfare, the response comes fast. Almost too fast.

“It’s not AI, it’s people.”

And as far as I know, these are often the same people who have defined AI as a socio-technical system meaning it’s not just code, but a network of actors: builders, deployers, policymakers, users.

The irony?

These are also the same people I’ve seen on major stages (e.g. AI For Good) two years in a row now, celebrating this framing while carefully avoiding the topic of AI in warfare. And when a real change-maker, Dr. Abeba Birhane, tried to bring this conversation forward, she was censored.

Let me be clear: Israel’s leadership is fully responsible for the decisions driving war crimes and its consequences on civilians. That accountability does not vanish because technology is involved. Nothing - absolutely nothing - in this piece shifts or reduces that responsibility. International courts and national mechanisms are doing their work on that front. What I am addressing here is not about replacing that accountability it’s about expanding it.

So why bring up guns?

Because we’ve seen this rhetorical move before. I’m not American, but I lived and worked in the U.S., and I’ve watched the gun debate up close. I’ve read the slogans. I’ve sat with the grief in headlines and cried over dead children who should still be alive - in the U.S., and now.. in Gaza. That grief doesn’t allow me to ignore the pattern when I see it.

And here it is again.

Saying “It’s not AI, it’s people” flattens a chain of responsibility that runs deep from commanders giving orders to engineers designing targeting algorithms, from military contractors deploying predictive models to Big Tech giants hosting the infrastructure that makes this all possible. I know it. I’ve seen it.

It sounds a lot like another familiar line:

“Guns don’t kill people; people kill people.”

And just like the gun debate, this rhetoric functions as a shield. It diverts the conversation away from systemic accountability from the design choices, profit structures, and governance failures that make harm scalable, faster, and harder to challenge.

The analogy isn’t perfect. Sure.

AI is not guns. Sure.

But it’s hard to unsee : the pattern is the same. Semantics are once again weaponized to block governance, normalize harm, and keep powerful actors out of reach.

Why This Phrase Feels Safe

When people say “It’s not AI, it’s people,” they think they’re reinforcing human agency. In reality, this line performs a political function: it makes the problem smaller.

It turns a global governance crisis into a simplified moral frame.

In the gun debate, that frame is individual morality: the lone shooter becomes the focus, while lobbying and access structures disappear.

In the AI warfare debate, it’s political culpability alone: state leaders become the sole focus, while the corporations building these systems and the governance spaces defining AI remain conveniently out of the picture.

And that has consequences:

• Governance remains focused on “safe” topics like bias audits and existential risk.

• AI in warfare stays out of scope untouched by standards, norms, or treaties.

• The hardest questions about militarization, profit, and power remain unasked.

Out of scope is not an accident. It’s a choice. It keeps governance comfortable, avoids tension with funders, and preserves the illusion of control.

And here’s the point, probably where I (personally) would like to step in: it delays the hard work of governance, our work.

Consider the gun debate “Guns don’t kill people; people kill people” became the NRA’s deflective slogan by the late 20th century. It reframed public conversation around shooters, not industry or access. As a result, meaningful regulation stalled. Even though the Gun Control Act of 1968 finally passed, it was “better than nothing” legislation, enacted after assassinations and mass protests, not because policymakers addressed the structural failure of weapon proliferation. The myth of neutrality made regulation appear unnecessary, so the industry stayed shielded.

That same rhetorical shield now protects the use of AI in warfare: once you insist it’s all about “people,” governance logic shrinks. Why regulate systems if the tool is neutral?

The Governance Blind Spot

I’ve been in AI governance for over a year, and this topic is treated as out of scope. It was even censored at AI for Good in Geneva. Two years of global AI summits, and I almost never hear the words warfare and accountability in the same sentence.

Meanwhile, the International Committee of the Red Cross (ICRC) has been warning about autonomy in weapons systems for years urging states to address humanitarian and legal risks. But those warnings barely enter the rooms where “AI governance” is discussed.

Why Naming AI Matters

Yes, we all know “AI” is an imperfect term. But let’s not pretend: this label is where traction, funding, and policymaking concentrate. Erasing AI from the warfare debate keeps it off the governance agenda.

It’s not about hype… or is it? Hype is a tool we didn’t choose, but it’s here, shaping agendas, unlocking billions, and deciding what gets regulated. If it’s powerful enough to dominate policy conversations on bias and existential risk, then it should be powerful enough to pull militarization into the governance room, for once, at good measure?

Using AI as an entry point isn’t so much about the hype that it is about accountability. If we don’t name it, we leave responsibility to others: the ICJ, civil society, fragmented human rights mechanisms.

Meanwhile, the so-called AI governance community gets to say, “That’s not our problem.”

A.D

** If you value this work, know that it takes hours of research and writing and it’s entirely unfunded, independent, and not backed by any institution or corporation. Consider subscribing or supporting to help keep these conversations possible.