Business responsibility in the Israeli-Palestine conflict: three tech companies, three different responses

Microsoft recently took the unprecedented decision of removing many senior executives of its Israeli operations and moving its reporting office from Dubai to France. 

The global software giant was stung by reports last year, which alleged that Microsoft’s cloud platform Azure was being used by the Israeli Defence Force to run a surveillance regime in the occupied territories and offer guidance to select bombing targets. Microsoft set up an investigation. Once the allegations surfaced, Microsoft hired a prominent law firm to investigate. 

The company discovered its Israeli subsidiary’s activities made it vulnerable to potential prosecution in Europe, which has strict data privacy laws, as well as undermined the company’s own human rights policies. Microsoft went on to block Israeli military units from using its technology. And now that it has evidence, it has taken firm steps to change the way it does business in Israel. 

Microsoft’s response is in marked contrast to the cavalier disregard Palantir has shown about its close ties with Israel or its technology and services being used to support Israeli attacks. Anthropic, another company offering artificial intelligence services, has drawn some lines, although some of its technology is used by Palantir and Microsoft, which links Anthropic’s services with conflict. 

Three cases, three different responses

No two technology companies are alike. But as modern warfare has come to rely more on technology, companies that provide technological hardware, software, and services, are no longer mere service providers offering logistical support. Their actions have real consequences on civilian lives. Many technological companies are involved with the conflict in the Middle East, and in three recent instances, three tech giants have reacted differently to the issues raised. The issues certainly affect reputations, but human rights is not about reputation management. They also raise legal risks, but large companies have access to legal services that can defend the corporations. What matters is whether the company adhere to international standards on human rights, comply with their responsibilities under human rights law, and operate within the framework of the UN Guiding Principles for business and human rights, and its recommendations for heightened due diligence in high risk environments and conflict-affected areas. Microsoft, Anthropic, and Palantir, three tech companies had similar warning signs: how they reacted offers lessons for the future. 

Microsoft first faced allegations in August 2025, when a joint investigation by The Guardian, +972 Magazine, and the Hebrew-language outlet Local Call threw light on an arrangement that had been operating in secrecy for years. 

The report showed how Israel's elite military intelligence branch, called Unit 8200, had been using Microsoft's cloud platform to build and run a mass surveillance system of extraordinary scale. The reporting alleged that the system was capable of collecting and storing recordings of millions of Palestinian phone calls every day from Gaza and the West Bank, which was internally described by its operators as "a million calls an hour." Three Unit 8200 sources confirmed to reporters that the data had been used to plan airstrikes, shape military operations, and justify arrests in the occupied territories.  

Deeper analysis showed that Microsoft headquarters did not have all the information about what was going on in Israel. Microsoft's engineers in Israel, some of whom included former Unit 8200 personnel, were working directly with the unit to tailor-make Azure to meet the Unit’s specifications, creating a segregated cloud zone with advanced security features. The Guardian further alleged that Microsoft provided an estimated 19,000 hours of engineering support worth $10 million to IDF intelligence units during the war. Many Microsoft employees protested within the company. 

Some Microsoft engineers apparently were aware the data included raw intelligence, while Israel-based staff, including Unit 8200 alumni, appeared to understand its military purpose. This had not surfaced in the initial internal review, which cleared Microsoft of wartime assistance. Senior executives were now confronted with the possibility that Israel-based staff had concealed information when questioned. The first review had cleared Microsoft partly because those who knew most told the least. 

The company commissioned a second, more rigorous external review led by law firm Covington & Burling LLP. On 25 September 2025, Microsoft announced it had ceased and disabled a set of services to Unit 8200, perhaps the first time a major US tech company revoked the Israeli army's access to any of its products since the start of its war on Gaza. The company probed deeper; a global team flew to Israel and found that certain units had been operating in ways that potentially violated the company's code of ethics. Most crucially, the Azure servers were based in Europe, which exposed Microsoft to regulatory exposure under EU law: this was a material risk that alarmed the company. 

Hence the leadership changes. General manager Alon Haimovich left, along with other senior staff in the subsidiary's governance department. Microsoft Israel, which reported to Dubai, was now placed under the management of Microsoft France. 

These steps are extremely meaningful. But it should be noted that Microsoft's actions were triggered not by internal criticism from its staff or whistle-blowing, but due to external journalism and subsequent legal review. Its other businesses in Israel are not affected (nor do they need to be, unless they raise the risk of complicity in acts considered criminal under international law). Amnesty International welcomed the Microsoft decision but called on the company to examine all its contracts, sales, and transfers of surveillance and AI equipment to Israel to ensure none are contributing to human rights violations.

Microsoft has since said it would enhance contract review processes, review controls for non-US markets, periodically review acceptable use of its products and services, provide guidance to employees, and provide mechanisms to employees to raise internal concerns.

Palantir: a sharp contrast

The behaviour of Palantir is in sharp contrast. It is proud of its ties with the IDF, although its chief executive admitted that his vocal support for Israeli operations in Palestine has led to employees leaving the company. Palantir has consistently maintained that reports of its involvement are inaccurate and some of its actions predate the conflict in Gaza. 

The company markets itself as a provider of Gotham, which it describes as ‘an AI-powered kill chain’ enabling decision dominance from space to mud. Its Maven Smart System uses computer vision and machine learning to scan satellite and drone feeds, flag potential targets, and signal appropriate weapons systems for kinetic strikes. Palantir agreed to a ‘strategic partnership’ with Israel in January 2024. Without specifying details, the company said it was giving the West ‘a critical edge’ in Middle East conflicts. The Pentagon, NATO, and the UK (where there are protests as Palantir is also working with the British National Health Service) have all signed contracts with Palantir.  

When challenged at a forum in May 2025 by a protester who accused the company of killing Palestinians, CEO Alex Karp responded: “Mostly terrorists. That’s true.” He has described Palantir's mission as being "to scare enemies, and on occasion, kill them." His candour has the virtue of clarity. But candour about complicity is not the same as accountability for it. 

The questions that international humanitarian law demands companies ask: whether they have leverage to ensure their partner acts consistently with the laws of armed conflict; and whether they have conducted any due diligence (including heightened due diligence during armed conflict) to determine if targeting systems comply with the obligation to distinguish combatants from civilians, appear not to have been asked at all.

Anthropic: a position of contradictions

A different kind of test was presented to Anthropic, whose AI model Claude was integrated into military workflows through Palantir's Maven platform. In July 2025, the US Department of Defence awarded Anthropic a $200 million contract, making Claude reportedly the most widely deployed frontier AI model across classified military networks.

Renegotiations broke down in February 2026 when the Pentagon demanded language authorising Claude for ‘any lawful use,’ which in Anthropic's reading would permit its deployment in fully autonomous weapons systems and for mass domestic surveillance. CEO Dario Amodei refused, stating that frontier AI is "simply not reliable enough to power fully autonomous weapons" and that the regulatory frameworks to ensure compliance with international law do not yet exist. The Trump administration responded by designating Anthropic a supply-chain risk to national security. This was the first time such a designation had ever been applied to an American company, effectively blacklisting it from military contracts. Anthropic sued (and Microsoft supported Anthropic with an amicus brief); a federal court issued a temporary restraining order. 

Anthropic's position is not without contradictions: it accepted a military contract, and its technology was used in lethal operations before the rupture occurred. But it drew specific lines against autonomous targeting without human authorisation, and against mass domestic surveillance, stated them publicly, and held them under extraordinary pressure at severe financial cost.

These three cases illustrate what it means for a technology company to take its own stated principles seriously, and what it looks like when it does not. The UN Guiding Principles on Business and Human Rights are clear: a company in a conflict area should view causing and contributing to gross violations of human rights as a legal compliance risk, and where a company lacks the leverage to prevent or mitigate adverse impacts, it should consider ending relationships with warring parties. The severity of harm determines the urgency of response. In a conflict where civilian casualties have included aid workers, journalists, hospitals, and UN staff, companies whose technology contributes to targeting decisions face questions of the utmost gravity. 

Microsoft’s stance was more consistent with due diligence expectations and its own standards which sets the bar for others to follow, not only for legal and reputational risks, but to ensure that the company’s conduct is consistent with international human rights law and standards. Anthropic has also lines and held them and suffered financial losses. Palantir has chosen, with open eyes, to build and sell the kill chain. The difference between them is not technical. It is moral.