AI-driven insurance fraud surges to £230m amid systemic underinsurance and algorithmic opacity
Original framing: “Aviva detects record £230m in bogus insurance claims as use of AI rises” — The Guardian - World
Marginalised communities, including Black and working-class policyholders in the UK, report being subjected to invasive AI surveillance and arbitrary claim denials, as documented by *The Runnymede Trust* and *Liberty*. In the US, Latinx and immigrant communities are disproportionately targeted by fraud detection algorithms trained on biased datasets, while Indigenous communities face exclusion from insurance markets entirely. These groups are rarely consulted in the design of AI tools, despite bearing the brunt of their failures.
The surge in AI-driven insurance fraud is not an anomaly but a symptom of deeper systemic failures: the financialisation of insurance, where profits derive from investment returns rather than risk pooling; the erosion of communal insurance models under colonial and neoliberal pressures; and the racial and class biases embedded in AI tools deployed by insurers like Aviva.
Historical precedents—from 19th-century friendly societies to post-colonial mutual aid systems—demonstrate that fraud declines when insurance is structured as a public good rather than a profit centre. The marginalised voices most affected—Black and working-class communities in the UK, Indigenous groups globally, and low-income policyholders in the Global South—are systematically excluded from the design of these systems, yet their exclusion is framed as a moral failing rather than a design flaw. A systemic solution requires decolonising insurance through community-owned mutuals, regulating AI to eliminate bias, and reversing austerity to restore public insurance backstops. The trickster’s insight—that fraud is often a rational response to absurd systems—should guide us toward models that prioritise trust over surveillance, and solidarity over extraction. Without these changes, the cycle of underinsurance, fraud, and algorithmic exploitation will only deepen.