EPO published patent application EP4639418A1 filed by Koninklijke Philips N.V. for 'Predicting the Reliability of Output Data.' The patent relates to AI reliability prediction methods and is classified under multiple IPC codes including G06N 3/084 (neural networks), G06N 5/01 (knowledge processing), G06N 7/01 (probabilistic models), and G06N 20/10/20 (machine learning). Inventors include Van Leeuwen, Vlutters, and Saeed. The patent application was published March 25, 2026.

Patent publications are informational filings that do not create immediate regulatory obligations. Companies developing AI reliability prediction systems should review the published patent claims to assess potential licensing needs or freedom-to-operate considerations. Patent protection, if granted, extends 20 years from the original filing date.