AXA Warns AI Cannot Replace Human Care as UK Mental Health Crisis Deepens

AXA’s Mind Health Report Reveals a Nation Turning to AI for Mental Wellbeing

More than one in three people in the UK are struggling with poor mental health, placing the country worst in Europe and third worst globally, according to AXA’s sixth annual Mind Health Report. With NHS waits for treatment stretching beyond 18 months and stigma still deterring professional help — 25% cited fear of judgement — nearly half of respondents (46%) have turned to artificial intelligence tools for mental health support.

The shift brings immediate consequences for the insurance industry. AXA UK & Ireland chief executive Tara Foley warned that the reliance on AI, which 42% of users trust more than a mental health professional, leaves a gap that technology cannot close on its own. The report found a stark split in experience: while 57% of AI users found the guidance helpful, 43% were dissatisfied, and 27% said an AI recommendation had led them into harmful behaviour.

That finding amplifies regulatory pressure already building around AI in financial services. The House of Commons Treasury Committee in January 2026 criticised UK watchdogs for failing to manage AI risks and urged the Financial Conduct Authority (FCA) to clarify how its vulnerable customer framework applies to AI-driven interactions. For health and protection insurers embedding AI into underwriting, claims and customer support, the data points to heightened scrutiny on outcomes for customers who may not self-identify as vulnerable.

The workplace dimension is equally direct. A companion survey of 2,000 UK adults found a third worried about AI’s impact on their career, 29% felt anxious about its use in their role, and 21% were considering a move to “future-proof” sectors. With group risk insurer intervention data from GRiD showing mental illness accounted for 48% of health and wellbeing cases and claims payouts hitting £2.69 billion in 2025, the report underscores a widening mismatch between demand for support and the capacity of public services — a gap that AI tools are filling, often with dangerous consequences.

What AI’s Failings Mean for Insurers, Employers and Regulators

Where This Leaves Insurers Building AI Into Products

AXA’s warning is not abstract. The 27% of AI users who reported harm from follow‑on advice creates a direct liability question for any insurer using AI tools in customer journeys — whether for symptom triage, mental health support apps or claims handling. If an insurer’s own AI delivers a similarly flawed recommendation, the reputational and regulatory fallout could be severe, particularly when dealing with vulnerable customers who may not flag their own distress. The FCA’s FG21/1 framework expects firms to identify and support vulnerability; an AI that inadvertently exacerbates a mental health condition would fail that test.

FCA Scrutiny of AI and Vulnerable Customers

The Treasury Committee’s criticism of regulators for failing to keep pace with AI risks in financial services is a clear signal that guidance and enforcement are likely to tighten. Insurers already building AI into health and protection products must now demonstrate that their tools neither replace necessary human oversight nor cause harm. The fact that nearly half of users trust AI more than a clinician — yet a significant minority report harm — suggests the technology’s reliability gap is a safety issue, not just a customer satisfaction one. If the FCA moves to mandate transparency or outcome testing for AI-driven advice, insurers with existing deployments could face costly re‑engineering.

The Widening Employer‑Insurer Gap

GRiD’s finding that 70% of employers believe state support for workforce health is insufficient — with mental health the most acute gap — places group risk insurers at the centre of a structural demand shift. As public services falter, employers will increasingly look to their group income protection and health insurance packages to deliver meaningful mental health intervention. AXA’s data that employees are anxious about AI in the workplace and are even switching sectors adds an extra dimension: insurers that can demonstrably combine fast access to human-led mental health services with safe, regulated AI supplements will be best positioned to retain and win corporate schemes. Those that over‑rely on unproven AI tools risk alienating the very workforces they aim to cover.

How Insurers Can Respond to AI-Driven Mental Health Risks

Businesses and insurers with exposure to mental health claims and customer vulnerability should consider the following steps, grounded in the report’s specific findings:

• Audit AI-assisted customer journeys for mental health risk. With 27% of AI mental health users reporting harmful behaviour after receiving advice, insurers must test whether their own chatbots, triage tools or claims interfaces could produce similar outcomes. Any tool that touches mental health conversations should include an immediate human escalation path and be stress‑tested against the FCA’s vulnerable customer principles.
• Prepare for tighter FCA guidance on AI in protection. The Treasury Committee’s January 2026 rebuke makes it highly likely the regulator will issue clearer expectations for AI handling of vulnerable customers. Insurers should start mapping their AI models against FG21/1 requirements and be ready to evidence that automated decisions do not worsen a customer’s mental state.
• Use group risk data to design products that bridge the mental health gap. With mental illness now responsible for 48% of group risk interventions and employer confidence in state support at a low, demand will grow for policies that offer fast access to human therapists — not just app‑based wellbeing tools. AXA’s model of giving health insurance members rapid access to clinical support offers one template, but the competitive advantage will lie in credible proof that human care, not AI alone, drives recovery.
• Address workplace AI anxiety or risk losing talent. The companion survey showed 21% of workers are considering a move to “future‑proof” sectors, largely driven by AI unease. For employers and group risk providers, offering mental health support specifically tied to AI‑related job stress — and communicating that support clearly — could become a retention differentiator.