How AI Is Reshaping Data Integrity in Pharma Manufacturing

With artificial intelligence (AI) now in pharma manufacturing, regulators have begun to pay close attention. In fact, oftentimes, they move forward without waiting for adequate guidelines. 

In the recent episode of The Chain Reaction podcast, Nour AlSaleh, Founder and CEO of Veragence Pharma Consultancy, sat down with host Shubhangi Dua, Podcast Producer and B2B Journalist at Pharmatica.io and EM360Tech.

AlSaleh shares her expertise on maintaining data integrity in this new era of AI and changing policies. She is a pharmacist with 20 years of experience in quality assurance roles at international and multinational pharmaceutical companies in Jordan. A few months ago, she started her own consultancy with a goal to help manufacturers in the Middle East and across the globe improve regulatory compliance and protect patient safety as digital and AI-driven systems transform the industry.

What Data Integrity Means Now in Pharma Manufacturing

AlSaleh describes data integrity as ensuring that every data point produced in a GxP or GMP environment, whether a sensor reading, batch record, or an AI model's output, is "complete, accurate, attributable, and trustworthy enough" to support safe patient decisions. 

However, this definition has had to expand, especially with technological developments. As AI tools are now amalgamated into manufacturing decisions, the Veragence CEO explains that the scope of data integrity has "widened to consider that we are no longer just talking about chromatography and paper-based documentation." 

Manufacturers must confirm that AI models receive correct data and that a human properly reviews every output before it affects a decision.

Read Maximise Pharma Innovation With Quality-by-Design

Maximise Pharma Innovation With Quality-by-Design

Find out how Quality-by-Design improves reliability and compliance in pharma and how you can maximise innovation through quality-based pharma development systems.

US FDA Warning Pharma Companies

AlSaleh points to a recent example where the US FDA sent a warning letter and a Form 483 observation to a pharmaceutical manufacturer after the company used an AI tool to establish product specifications without sufficient human review.

According to the U.S. FDA, an FDA Form 483 is a document issued to a company's management at the conclusion of an FDA facility inspection. 

According to the CEO, companies cannot use regulatory uncertainty as an excuse for inaction. "We cannot have this justification that we still don't have clear guidelines from the regulators," she states. 

"Patient safety is not negotiable." She believes AI should act like a co-worker—helpful for speeding up analysis and identifying trends, but never a substitute for human judgment. "It will never be a replacement for the human brain."

Why ALCOA+ Needs Reinterpreting for Pharma AI

The podcast also ventures into ALCOA and its extended version, ALCOA+. ALCOA is a data integrity framework that combines five fundamental principles used to ensure data integrity, compliance, and quality. It stands for attributable, legible, contemporaneous, original, and accurate. 

ALSaleh believes that these frameworks still hold, but each principle now needs a fresh perspective. For instance, regarding "attributable", if an AI model logs a deviation, who is actually responsible? "We need to specify who is accountable in this case. Is it the user, the data scientist, or the vendor?"

She mentions that EU regulators have begun drafting Annex 22 specifically to address AI in manufacturing, but global consistency is still lacking. Manufacturers operating in multiple markets—the US, EU, Australia, Canada—often must interpret AI compliance expectations on their own.

EU GMP Annex 22 is the first dedicated regulatory framework governing the use of Artificial Intelligence (AI) and Machine Learning (ML) in pharmaceutical and medicinal manufacturing according to Eupry. 

Throughout the conversation with Dua, AlSaleh also covered where data integrity breaks down and the associated risks. She also challenges the notion that AI and "vibe coding" methods can bypass compliance in pharma.

In terms of the future, the Veragence Founder foresees more regulatory discussions with manufacturers, greater harmonisation of global AI guidelines, and an increasing need for talent that combines GxP regulatory knowledge with AI skills. 

Her advice to pharma leaders is to invest in that talent now, encourage open dialogue with regulators early in development, and always prioritise human oversight over speed.

AlSaleh emphasises that the goal is not to choose between fast innovation and compliance. A solid compliance foundation enables quicker, safer innovation.

Listen to the full episode of "The Chain Reaction" by Pharmatica.io for the complete conversation with Nour Al Saleh.

Takeaways

• Data integrity is foundational for product quality and patient safety.
• AI and digital systems are reshaping the pharmaceutical industry.
• Human oversight is essential in evaluating AI-generated data.
• Regulatory expectations are evolving with the introduction of AI.
• The interface between systems is a common risk for data integrity.
• Outsourcing does not absolve manufacturers of responsibility.
• Continuous monitoring of AI tools is necessary for compliance.
• Data integrity requirements should be integrated from the design phase.
• Knowledge and training are crucial for effective data integrity control.
• Patient safety must remain the primary focus in pharmaceutical manufacturing.

Chapters

• 00:00 AI and data integrity overview
• 02:19 Noor’s background in pharma
• 03:45 Defining data integrity today
• 05:33 Leader responsibility in compliance
• 07:30 Alcova Plus Plus in practice
• 09:41 Regulatory expectations for AI
• 12:16 FDA warning letter example
• 14:14 Risks at system interfaces
• 16:08 Vendor oversight and lifecycle control
• 18:55 CSV for evolving AI tools
• 20:23 Drift, monitoring, and revalidation
• 23:20 Design-phase data integrity
• 25:00 Choosing the right vendor
• 27:36 Traceability in automated systems
• 31:32 Inspection readiness culture
• 35:17 AI as a pharma partner
• 40:09 Priorities for future pharma leaders
• 43:29 Closing thoughts

For more information on data integrity and AI in pharma, please visit veragence.com.