A Manufacturing Defect Fed Bad Data to Abbott's AI Glucose Monitors and Seven People Died

What happened

In November 2025, Abbott recalled approximately three million continuous glucose monitoring sensors after regulators and patients connected 736 severe adverse events and seven deaths to inaccurate readings from the devices. The affected products, FreeStyle Libre 3 and FreeStyle Libre 3 Plus, are AI-powered sensors worn by people with diabetes to automatically track blood glucose without routine finger-stick testing. Abbott described the action as a "medical device correction," but the seven deaths reported before the correction was announced tell a more serious story.

The sensors work by collecting raw glucose data from interstitial fluid and passing that data to an AI system that infers blood glucose levels, identifies trends, and issues alerts when levels fall dangerously low. For people managing insulin-dependent diabetes, those alerts are not a convenience; they are the mechanism by which a patient knows whether to eat, whether to inject, and whether to seek emergency care. When the readings are wrong, the patient has no independent signal to check against, and the system's confidence in its own output can make the error invisible.

The root cause was a manufacturing defect traced to a single production line. That defect corrupted the physical sensor, which then fed inaccurate data to the AI layer processing it. The AI produced bad outputs from those bad inputs: delayed alerts, incorrect glucose readings, missed warning signals for hypoglycaemia or hyperglycaemia. Patients who trusted those readings administered wrong insulin doses or failed to act on a dangerous glucose episode the sensor did not register. The AI component did not introduce the flaw; it amplified the consequences of one.

The timeline raises harder questions than the defect itself. Abbott stated the manufacturing flaw had been identified and resolved, and the company offered free replacement sensors. But 736 severe adverse events and seven deaths were reported globally before the public correction announcement, across multiple countries where the sensors were in active use. Whether all affected users received timely notice, and whether the safety signal was escalated as quickly as the eventual recall statement implies, is not settled by the recall announcement alone.

Medical devices that incorporate AI add a specific accountability problem to the standard product-liability question. A manufacturing flaw in a traditional device is traceable through physical inspection and batch records. When the flawed component feeds into a predictive system, the failure path runs through software outputs that patients and clinicians can only observe, not audit. A provable record of what a system did, what inputs it received, and when an anomaly first appeared in aggregate device telemetry is the infrastructure that separates a timely correction from a recall that arrives after the body count is already tallied.

Reported impact

Affected parties

Not publicly disclosed

Harm type

Not publicly disclosed

Scale

Not publicly disclosed

Financial impact

Not publicly disclosed

Regulatory action

Not publicly disclosed

Classification

Organization

Not publicly disclosed

AI system

Not publicly disclosed

Industry

Not publicly disclosed

Country

Not publicly disclosed

Provider

Not publicly disclosed

Incident type

Not publicly disclosed

Relevant governance controls

Governance control mapping is not available for this record.

• No controls mappedNot publicly disclosed

Control mapping is analytical. It does not state that any control would have prevented the incident.

Sources and evidence

This record was researched and written by the Index. The event is also catalogued in the following database, which is listed for cross-reference.