Google Might Be Developing AI Weapons for the Next AI-Powered World War

Google just crossed a line no one thought they would. Quietly erasing their promise never to build AI for weapons or surveillance. With billions pouring into AI, military deals in the shadows, and a global arms race accelerating, we might be watching the most dangerous shift in technology history unfold right now. Let's rewind for a second. Back in 2018, Google published this big list of AI principles. Right after the whole controversy with Project Maven, which was a Department of Defense program that used AI to analyze drone footage, a bunch of Google employees protested, saying they didn't want their work to contribute to weapons that might harm people. In fact, some employees resigned and thousands signed a petition because they felt the AI work they were doing was crossing an ethical red line. And Google then decided not to renew that Pentagon contract and made a public promise that that they would not design or deploy AI for purposes like weapons or surveillance that violated internationally accepted norms. They were taking a stand. Well, as of February 2025, that promise is gone. Now this announcement came on February 5, 2025, when Google updated what they call their AI principles. In its new policy. Google basically says that it's still going to develop AI responsibly and in line with widely accepted principles of international law and human rights. But they've totally removed that specific commitment not to build AI for weapons or surveillance. Instead, DeepMind chief Demis Hassabis and Google Research Labs SVP James Manika wrote in a blog post that there's this increasingly complex geopolitical landscape and so they believe that democracies should be leading the race in AI guided by freedom, equality, respect for human rights, and so on. They also said that that collaboration between companies, governments and organizations is key to protecting people and supporting national security. So you can probably see the shift. They're prioritizing national security and competition with other countries over that earlier vow not to develop certain types of AI. Interestingly, this updated policy drop comes not long after Google's parent company Alphabet, reported slightly disappointing earnings. Their revenue was about $96.5 billion, which was just under analysts expectations of $96.67 billion. That announcement caused Alphabet's shares to tumble by around 8% when Wall street opened. A senior analyst at Emarketer, Evelyn Mitchell Wolf, mentioned that Google Cloud's slower than expected growth was a big part of why the numbers came in lower. She also suggested that the AI powered momentum might be losing steam, especially since there are rising questions about Google's more closed model strategy. And there's even mention of a competitor called deepseek that embarrassed Google's AI on the World stage Speaking of big spending, Alphabet says it'll pour $75 billion into capital expenditure next year, mostly to build out AI capabilities and infrastructure software. All in all, Google is going all in on AI, even if that means changing its original stance on weapons and surveillance. Now, to give you a sense of the broader picture, Google has had a kind of complicated relationship with this don't be Evil motto. From its early days that was originally the company's big moral standard, but by 2009 it'd been downgraded to a mere mantra. And when Alphabet was created in 2015 as Google's parent company, don't be evil wasn't included in the new code of ethics. So maybe this new pivot away from disclaimers about not building AI weapons is in line with Google's older pattern of moving away from simpler, more black and white moral statements. Meanwhile, a lot of employees are responding to all this in real time. Internally. They have a message board called memejin, and apparently there are memes floating around, one referencing that comedic Are we the Baddies Nazis? Sketch and another with Sheldon from the Big Bang Theory going, oh, that's why. After seeing reports of these new military friendly partnerships, another meme jokes about CEO Sundar Pichai googling how to become weapons contractor. Some employees, of course, might be okay with aligning with defense and government work, seeing it as necessary or even patriotic, especially if it's about strengthening national security or protecting troops on the ground, as some put it. Google has over 180,000 employees, so it's a big population with a range of opinions. And speaking of opinions, Andrew Ng, who founded Google Brain and was a key figure in shaping the company's AI initiatives early on, says he's very glad Google changed its stance and dropped the old promise. Ng spoke at a military veterans startup conference in San Francisco on February 7, 2025, and basically said he never understood why so many employees were upset over things like Project Maven. He said something along the lines of if service members are willing to shed blood for our country, how can an American company refuse to help them? And he also expressed relief that certain AI regulations, like California's SB 1477 bill and President Biden's former AI executive order got overturned because he believes those measures would have slowed down American AI innovation, giving other countries a leg up. He's definitely in the camp that wants the US to maintain an advantage in AI, especially over China. He even mentioned how AI drones could transform the battlefield. Another former Google executive, Eric Schmidt, has been pushing A similar message in Washington advocating for the government to purchase AI drones to compete with China. Of course, not everyone feels that way. Meredith Whitaker says, who led those 2018 protests at Google, is very much against developing AI for warfare. She said at the time that Google should not be in the business of war. And Nobel laureate Geoffrey Hinton, another major AI researcher, has called for governments worldwide to limit or ban AI in weapons. Jeff Dean, now Google DeepMind's chief scientist, once signed a letter that opposed using machine learning for lethal autonomous weapons. So there's a lot of division within the AI community now, if that isn't enough to keep track of OpenAI has also stepped into the spotlight with a massive new partnership with the US Government. The plan is for the national laboratories, places where up to 15,000 scientists work on nuclear research, to use OpenAI's latest O1 series of models to help secure nuclear weapons and materials. That alone sounds like something straight out of a sci fi movie. Sam Altman, OpenAI's CEO, says they want to focus on reducing the risk of nuclear war, but many people are worried about letting an AI that's known to hallucinate or leak private info anywhere near nuclear secrets. People are referencing the Terminator half jokingly, half seriously, because in that movie an AI defense network basically decides to wipe out humanity. So it's not hard to see why folks might be a bit nervous. Adding to the swirl of controversy is the fact that Sam Altman attended President Donald Trump's January 20th inauguration this year, joining tech leaders like Jeff Bezos and Mark Zuckerberg. Trump, by the way, wasted no time after taking office to rescind a former Biden executive order that mandated companies share results of AI safety tests with the government. That means there are fewer guardrails in place now. Also, Altman apparently donated $1 million to Trump's inauguration and publicly said that Trump has really changed my perspective on him, which surprised a lot of people because Altman used to be critical of Trump. And if you've heard about the new $500 billion Stargate venture the Trump administration introduced for building AI infrastructure, well, OpenAI is going to invest tens of billions into that as well. All this is happening while OpenAI is reportedly in talks for yet another big round of funding that could value the company at a whopping $340 billion, double its previous valuation. It's no wonder they're looking to expand their influence in government sectors. Plus, they announced something called chatgpt.gov, aimed specifically at US government agencies focusing on security. Exactly how bulletproof that security is remains to be seen, especially given the track record of these models making stuff up or revealing user data inadvertently. One of the biggest reasons behind these moves is the new AI arms race between global superpowers. China has invested heavily in AI, and there's mention of a Chinese startup named Deepseek that recently showcased it can compete with the best models out there. Google's blog post even says that democracies need to take the lead in AI to ensure it develops in a way consistent with human rights. Obviously, that's not a small concern, but whether that means working on weapons AI is the best method is still a matter of fierce debate anyway. Google itself has engaged in multiple defense contracts over the years, even before the new change in policy. Project Nimbus, for example, is a partnership with the Israeli government to provide cloud services. There were also internal protests about that, with employees saying it undermined Palestinian rights. Meanwhile, Amazon is working with Palantir on something that enables the Claude AI for US military and intelligence customers. So Google's not alone. Microsoft, Amazon, Meta, and other major players are also wrestling with the ethics of selling AI to militaries. Stuart Russell, a British computer scientist, has delivered Reith lectures warning about autonomous weapons systems. He wants global controls. This new environment, though, is basically the opposite. Companies are removing constraints. Google, in the same breath, says they want to stay responsible. But for a lot of people, that begs the question of what responsible even means if you're actively building, or at least leaving the door open to building lethal tech. Google quietly dropped its promise not to use AI for weapons or surveillance, sparking mixed reactions. Some see it as necessary for national security, while others call it a betrayal of ethics. With billions pouring into AI, growing military ties, and fierce competition from rivals like OpenAI and Deepseek, the line between innovation and militarization is getting blurrier than ever. Thanks for watching, and I'll catch you in the next one.