Paper forms are still the primary data capture mechanism for a significant portion of GxP-relevant activities in pharmaceutical manufacturing, clinical operations, and quality management. Batch records, environmental monitoring logs, equipment cleaning records, deviation reports, in many facilities, these are captured on paper, signed by hand, and filed in binders.

This approach has a compliance cost that has always existed. In 2026, it has an additional cost: paper data cannot feed AI systems. Organizations that want to use AI for quality improvement, anomaly detection, or predictive analytics have to start with structured, digital, traceable records. Paper forms produce none of these.

Every pharmaceutical organization we speak with wants to use AI. Anomaly detection. Predictive quality. AI-accelerated validation. GenAI agents for deviation support and SOP guidance. Almost none of them have the data foundation required to do any of this at scale in a GxP-compliant way.

This is not a technology problem. The AI models exist. The regulatory guidance, FDA's 2025 AI Draft Guidance, Annex 22 from EMA, the 2026 CSA Guidance, is increasingly clear. The problem is the data layer underneath the AI. And that layer has three structural gaps that most organizations have not yet addressed.

Data integrity programs in pharmaceutical organizations typically focus on what people do with data: how forms are completed, how records are approved, how audit trails are maintained in validated systems. These are the right things to focus on. But they miss a significant and often overlooked source of risk: what lab instruments do with data automatically, and where that data goes after the instrument generates it.

Lab instruments, chromatography systems, spectrophotometers, balances, dissolution testers, generate output files continuously during normal operation. These files contain the raw data that underlies testing decisions, batch release determinations, and stability conclusions. They are the most critical GxP data in the organization. And in most pharmaceutical facilities, they sit in uncontrolled network folders with no access restrictions, no audit trail, and no version management.

Spreadsheets are everywhere in pharmaceutical and life sciences organizations. Clinical trial data, raw material testing results, manufacturing logs, assay calculations, training records, the list goes on. Studies consistently show that spreadsheets are used to support GxP-critical processes at most pharma and biotech companies, often without formal controls, validated audit trails, or access restrictions.

That has always been a compliance risk. In 2026, it is also an AI readiness problem. And the two are converging at exactly the moment FDA enforcement is intensifying.

Spreadsheets, lab data, paper records, and AI: what the regulation requires, where organizations fail, and how to fix it.