AI is being used in warfare, and people are not happy about it, at least four million people to be exact. The #QuitGPT movement has been gaining momentum since the start of January, with 700,000 users pledging to cancel subscriptions in February alone.
A QuitGPT spokesperson told The New Arab: “In five weeks, we got over 200 million views on our social media posts that we released earlier this year.” With over four million signing the petition to QuitGPT, the movement estimates it has decreased parent company OpenAI’s market capitalisation by $1.5 billion.
The movement emerged as news surfaced showing the intrinsic link between OpenAI and the Pentagon as it pivoted toward becoming a primary defence contractor, in addition to when, OpenAI president Greg Brockman and his wife donated $25 million to Trump’s MAGA Inc in 2025.
Jovana Davidovic, assistant professor of philosophy at the University of IOWA and chief ethics officer at BABL AI, told The New Arab: “It’s not necessarily about OpenAI as a company, it’s about the use of LLMs [Large Language Models] in warfare, and the issue of mass surveillance.”
The typical AI chatbot that operates via LLMs is not a weapon of war on its own. Yet it can form an integral part of a military system. LLMs can be used in war settings to summarise intelligence, rank high-priority threats, and recommend targets.
The issue is that these processes can be made with minimal human input. The expert noted that LLMs “pose significant risks for warfighting, in terms of safety, escalating conflict, and risks to the attribution of responsibility.” Civilians are at risk, as are the forces operating LLMs; it is not only about ethics but also “an issue of operational safety,” Davidovic said.
The current US-Israeli attacks on Iran have enhanced the conversation around the digital non-alignment movement, as headlines dubbed it “the first AI war.”
Yet AI is not a new tool in warfare. What makes the emerging story about AI in warfare different now is “the use of LLMs, and these are a new beast that bring a different type of risks and new worries,” the professor explained.
AI warfare in the Global Majority
Out of the 28 regional conflicts, 27 are in the Global Majority (Global South), according to the Council on Foreign Relations tracker. The only country with an active conflict in the Global North is Ukraine.
One of the QuitGPT movement’s viral posts released in February was on workers in Kenya, based on a DW report, who have to sit watching explicit, abusive content or have disturbing conversations for hours on end with AI to manually train LLMs.
The spokesperson reflected on the heartbreaking irony in the situation. “My understanding is that there are many different countries in the Global South who are employed to do this kind of thing to train a model that could ultimately be used to bomb their own country.”
The US-Israeli war on the Arab region is one key example of AI in warfare. Notably, the US Department of War, formally the DoD, has used AI to identify targets across Iran, Iraq, Syria, and Yemen. Most recently, Operation Epic Fury hit 1,000 targets in Iran within 24 hours.
Davidovic commented on the quantity of the targets and the time frame, saying, “1,000 targets in 24 hours, that’s a lot, you can’t put human eyes on every piece of data that underpins that amount of target nomination.”
The quantity of data AI produces can pose serious problems in this regard. “AI enables such large numbers of target nominations and target identifications that humans can’t provide meaningful oversight for that,” Davidovic noted.
What is clear is the way AI is changing the way everyone sees the world, from our interactions to whether a conflict comes to fruition. “AI is one of the major driving forces behind what is happening when it comes to conflict,” the expert noted.
Most LLMs are owned by private companies, and Davidovic named this as one of the primary issues. “What does that mean for the safety of our data and the safety of our military data?” The data gathered by LLM is then being shared with big actors, which Davidovic noted “are usually within the US or China. I think that’s one big worry on everybody’s mind, and it should be.”
From BDS to QuitGPT
Many resistance movements have spiked over recent years, such as the Boycott, Divest, Sanction (BDS) movement and the university encampments to target companies that support Israel amid its genocide on Palestine.
The QuitGPT spokesperson discussed the similarities between the movements and the QuitGPT uprising, saying: “I think there are strong parallels between them; people are furious, so they take action.”
The movement is supported by students from several universities. “Students have been reaching out to the movement and saying to start petitions to get their school to divest from OpenAI,” the spokesperson explained, noting this is very similar to the encampment demands.
The age of ChatGPT users is highly relevant, as young people make up a large share of AI users — meaning the momentum behind QuitGPT among younger generations could have a severe impact on industry giants. In the US alone, estimates suggest that in 2024, around 59% of 18 to 24-year-olds use ChatGPT, and 56% in the 25-34 age bracket.
ChatGPT uninstalls in the US alone, increased by 295% on 28 February, as consumers responded to the news of OpenAI’s deal with the Department of War. “It is not possible to say exactly that we as the movement played a part in that, but it seems very significant,” the QuitGPT spokesperson highlighted.
Does ethical AI exist?
Anthropic, the owner of Claude, made headlines in February when it said that it would not back down in a fight with the US Department of War over how its AI technology is used.
Of all the publicly available AI platforms, is there one that is 100% ethical? “I don’t think so,” Davidovic said. She added that Anthropic “seems to be trying to lead the way on being explicit about their guardrails in terms of military use, but there are just constant new risks.”
The stance Anthropic has taken matters, Davidovic highlighted, because they are standing up and saying, “We know the bad and good way to use this technology, and how to make sure the use is compatible with the Constitution.”
Although there have been conversations in the international community on best practices and principles, she explained, “It’s important that we set norms, and we don’t have international norms yet; there is no consensus on how we want to use this cyber technology when it comes to killing people.”
From QuitGPT’s perspective, there is “absolutely a desire” for ethical AI tools.
The spokesperson concluded, “It’s about having access to that type of functionality without feeling like you need to morally compromise yourself or support a company that is engaging in things that you don’t believe in.”
Olivia Hooper is a British journalist based in Morocco, specialising in humanitarian and gender-based topics, and is the co-director of Politics4Her
