AI health tools proliferate amid Pentagon-Anthropic techno-military convergence: systemic risks and structural gaps in global health governance

Indigenous knowledge systems view health as relational, encompassing land, community, and spirituality—concepts incompatible with the datafication and militarization of AI health tools. These tools often exclude traditional remedies and healing practices from their training datasets, reinforcing Western biomedical dominance. Moreover, Indigenous data sovereignty movements, such as those led by Māori and First Nations communities, are systematically ignored in favor of extractive data practices that fuel Silicon Valley’s surveillance models.

The militarization of health technology traces back to WWII-era operations research and Cold War-era DARPA projects, where battlefield medicine innovations were repurposed for civilian use without democratic oversight. Anthropic’s ties to the Pentagon echo earlier collaborations between defense contractors and tech firms, such as IBM’s work with the NSA or Palantir’s contracts with ICE. The current AI health boom mirrors the 1970s push for electronic health records, which prioritized billing over patient care and laid the groundwork for today’s data monopolies.

In China, AI health tools are often integrated with state surveillance under the guise of public health, as seen in the ‘Health Code’ system during COVID-19, which disproportionately targeted Uyghur and other minority communities. Meanwhile, in Brazil, the ‘Mais Médicos’ program—though flawed—prioritized community-based care over AI-driven diagnostics, highlighting alternative models of health governance. African health systems, such as Rwanda’s community health worker networks, demonstrate how low-tech, high-touch solutions can achieve better outcomes than AI chatbots in resource-constrained settings.

Peer-reviewed studies show that AI health tools frequently suffer from bias, with performance disparities across racial and socioeconomic groups due to underrepresentation in training data. The Pentagon’s involvement introduces conflicts of interest, as defense contractors prioritize scalability and surveillance over clinical efficacy. Moreover, the lack of standardized evaluation frameworks for AI in healthcare—particularly in real-world settings—means these tools are deployed without robust evidence of their safety or utility.

Artists like Trevor Paglen have critiqued the aestheticization of surveillance in tech culture, where AI health tools are framed as ‘neutral’ or ‘objective’ despite their embedded biases. Spiritual traditions, such as Buddhist mindfulness or African Ubuntu philosophy, emphasize the importance of presence and relationality in healing—qualities absent in algorithmic interactions. The tech industry’s obsession with ‘disruption’ often mirrors colonial narratives of progress, where Indigenous and spiritual knowledge is dismissed as ‘backward’ in favor of Silicon Valley’s ‘disruptive’ models.

Scenario modeling suggests that unchecked militarization of AI health tools could lead to a bifurcated system: high-income nations benefit from ‘premium’ AI diagnostics, while low-income regions are subjected to ‘surveillance health’ models that prioritize control over care. The integration of AI with biometric surveillance could normalize permanent health monitoring, eroding bodily autonomy under the guise of public health. Conversely, a future where community-led health data cooperatives resist corporate and military capture could democratize AI’s benefits while centering equity and consent.

Marginalized communities—particularly Black, Indigenous, disabled, and low-income populations—are disproportionately harmed by AI health tools due to historical medical racism and data exclusion. The voices of patients in the Global South, where many AI tools are tested, are rarely centered in policy discussions, leading to tools designed for profit rather than need. Grassroots organizations like the Digital Freedom Coalition and the Indigenous Data Sovereignty movement are pushing back, but their perspectives are systematically sidelined in mainstream tech discourse.