Together with Privacy International we wrote a report on the role of companies in the integration of AI in warfare and the imminent threat it poses to civilians. Read the full report here.
Rapid rise of artificial intelligence
Information technologies have become crucial in the development of new weapons and in warfare. This military revolution is built on anything from advanced microchips and cloud services to AI generated targets and autonomous drones. We can witness this on an almost daily basis, from Ukraine to across the Middle East.
Amid booming military budgets and unprecedented investments in AI, global tech and arms industries are being reshaped, where tech giants have become key suppliers to the military, and traditional arms producers race to build increasingly autonomous weapons.
Why is this important?
Over the past decade we have seen how a dozen of mainly US tech companies grew to become the world’s largest companies. The originally consumer-oriented companies such as Google, Amazon, Meta and Microsoft have become heavily invested in the military domain. This goes partly through offering dedicated military cloud solutions, partly through co-operating with arms producers in the integration of their technologies into weapon systems.
Especially since the start of the second term of the Trump Administration, the biggest players in the sector appear to have fully embraced the military applications of their technologies now that their access to the presidency is closer than ever before and US government has now put AI front and centre of their policies. Also, the eight richest people on the planet – all men – are connected to the world’s biggest tech companies. Especially when democracy and the rule of law risk being compromised this is a toxic combination. US Secretary of War Hegseth for example advocates “maximum lethality” for US forces and has expressed disdain for what he called “stupid rules of engagement” designed to minimize civilian harm.
The current acceleration of technological progress and the level of geopolitical tension and conflict demand a much more urgent response from states and the United Nations, to ensure that ethical principles and international law are upheld to protect civilians against the use of new technologies.
Tech companies
In recent years some of the largest tech companies have signed multi-billion military contracts, as part of a trend where civil and military technologies become increasingly intertwined.
Nvidia, the world’s most valued company, sells chips that drive the most powerful data centres, and are also found in both Russian drones and American fighter jets. Nvidia is also the largest tech employer in Israel and works with Israel’s largest arms producer Elbit. Palantir and Lockheed Martin are other key military business partners. Clearly, the company is directly linked to arms companies that have enabled atrocities among civilians, from Gaza to Iran.
Alphabet, Amazon, Microsoft and Oracle won the Pentagon’s USD 9 billion Joint Warfighting Cloud Capability contract in 2022. The same year, the National Security Agency (NSA) awarded Amazon a cloud computing contract codenamed “Wild and Stormy”, worth up to USD 10 billion. Two years earlier, Alphabet, Amazon, IBM, Microsoft and Oracle won a contract to supply cloud services to the US intelligence community; likely worth “tens of billions” USD over 15 years. These are the largest such tech contracts to date.
More recently, tech companies are now heavily invested in generative AI, either by developing models or through major stakes in relatively new companies such as OpenAI (ChatGPT) and Anthropic (Claude). US ‘GenAI.mil’ contracts for Alphabet, OpenAI and SpaceX are expected to grow significantly in the coming years, as part of the aggressive US AI Strategy.
Anthropic was initially part of the deal but was dropped by the Pentagon after it demanded that its programmes would not be used for internal surveillance or lethal autonomous weapons. Even so, the use of generative AI in US offensive military operations in Venezuela and Iran, particularly Anthropic’s Claude, has been reported in recent months.
Facebook parent company Meta only recently joined the military circuit but has quickly embraced cooperation with some the most controversial players in the field, including Scale AI, Anduril and Palantir; Microsoft too cooperates with Anduril and Palantir.
Alphabet, Amazon, Microsoft and Oracle have long and extensive relations with the Israeli armed forces, which deepened further after the start of the ongoing genocide in Gaza in October 2023. Amazon and Google for example provide extensive cloud services to the Israeli government and the IDF, under the USD 1.2 billion Project Nimbus. Microsoft for its part is said to have a “footprint in all major military infrastructures” in Israel.
The relatively new players Anduril and Palantir have become emblematic of a US drive towards increasingly automated warfare with few constraints; their AI-enabled Maven Smart System is one of the flagship examples. Both work closely together and have successfully challenged the position of legacy arms-industry players. Anduril is rapidly expanding the range of weapons it offers, including advanced autonomous technologies such as loitering munitions and collaborative combat aircraft (CCA).
International guidelines
Current military uses of AI – both in providing targeting suggestions and in increasingly autonomous weapons – are set to expand and proliferate. This makes the call for new binding international rules aimed at responsible, reliable and accountable use of information technologies in military applications more urgent than ever. Such rules must ensure human agency and reaffirm international law. Without clear political will to set up guardrails, an AI arms race can spiral out of control and put the whole world at risk.
After more than twelve years of discussion, states have a unique opportunity later in 2026 to decide to open UN negotiations on a treaty on autonomous weapons. Moreover, a separate track of UN General Assembly mandated discussions on the broader issue of ‘AI in the military domain’ is taking place, with informal exchanges starting in June 2026.
Which actions and whose responsibility?
States bear the primary responsibility to create, adapt and enforce international rules for a changing world, but companies, investors and financial institutions have a key role and responsibility too. They need to ensure that their products comply with international norms on business and human rights, including human rights due diligence to ensure their products and services do not contribute to violations.
Together with Privacy International* we therefore call on…
… states to:
• start negotiations on a treaty on autonomous weapons without delay
• adopt a moratorium on the use of AI systems for the use of force, for example in decision support systems, until necessary international rules and effective safeguards are in place;
… companies involved in AI used for military purposes to:
• stop developing or selling autonomous weapon systems that operate without meaningful human control, and stop supplying AI systems for the use of force, until necessary international rules and effective safeguards are in place;
• establish clear public policy committing not to contribute to the development, production or sale of such systems;
… investors and financial institutions to:
• adopt investment and financing policies on AI in the military domain to address and mitigate human rights risk posed by these technologies; and
• require the companies they invest in or finance to guarantee that their activities do not contribute to the development, sale, or transfer of autonomous weapon systems without meaningful human control, and AI systems for the use of force.
Together with Privacy International, we investigated how AI and other digital technologies are rapidly integrated into warfare, how and which companies play a key role and how international law relates to these developments. The full report explores this theme in the light of the current geopolitical context and conflicts. It can be read here.
