Signal 48% of executives fear AI is injecting misinformation into their own datasets

Summary

Nearly half of executives fear that AI could be introducing false information straight into their own company data. Deloitte's 2026 survey puts the figure at 48 percent. The concern is that small errors in these datasets can spread and grow into serious operational and ethical failures. Deloitte observes that organisations are increasingly aware of external misinformation, such as deepfake fraud, but often neglect the quality of internal data. It recommends AI lineage mapping, which tracks where data comes from and how it is transformed and used across training and inference. Technologies such as blockchain can add tamper-proof, time-stamped records to strengthen that lineage.