While AI is taking off in biopharma for administrative tasks and drug discovery efforts, more than 90% of drugmakers have yet to commit to scaling AI use in regulated biopharma manufacturing, according to CRB’s new Horizons: Life Sciences report.
Fewer than 10% of drugmakers are scaling AI to support clinical or commercial manufacturing, a survey of more than 400 R&D and manufacturing professionals found.
CRB, a company that designs and builds biopharma facilities, ran the survey for its 2026 Horizons report on the investment choices of life sciences companies. The survey found widespread use of AI in nonregulated areas, with 56% of respondents applying the technology to administrative workflows and 45% using it in drug discovery. Yet uptake in regulated manufacturing is lower.
Asked about their use of AI in clinical or commercial manufacturing, just 1% of respondents said that they are currently using the technology in most or all applications, while 8% are scaling AI for use in many regulated manufacturing applications and 17% are scaling the tech for a few applications.
The remaining 74% of respondents could be several years away, at best, from applying AI to regulated medicine production.
CRB linked the figures to concerns about the suitability of current AI models for regulated manufacturing environments. Almost half of respondents named accuracy or quality concerns as major issues for the use of AI in clinical or commercial manufacturing. Insufficient validation standards were the next most commonly cited issues, followed by unclear accountability and a lack of transparency and explainability.
Respondents did appear to see an AI-based future. CRB found that 31% of those surveyed, including 24% of people who have yet to secure funding for AI adoption, plan to use the technology in the next three years. Another 31% of respondents are considering using AI in drug production but are yet to adopt a formal plan for the rollout. Just 12% of respondents have no plans to use AI in regulated manufacturing.
The findings echo comments by manufacturing leaders, who have noted the challenge of providing the interpretable outputs expected by agencies when using models with internal workings that are a mystery to users. Experts have also reported data integrity and traceability issues linked to the need for models to be version-controlled, audit-trailed and compliant with FDA regulations governing electronic records.
Indeed, AI use triggered a warning letter from the agency in April. FDA inspectors found Purolea Cosmetics Lab’s use of AI to create drug product specifications, procedures and master production or control records violated rules on the responsibilities of quality control units. The FDA told Purolea that “any output or recommendations from an AI agent must be reviewed and cleared by an authorized human representative.”
