Current U.S. surveillance for highly pathogenic avian influenza A(H5N1) in humans prioritizes individuals with known animal exposures, potentially missing community-acquired infections. To address this gap, we implemented universal H5 subtyping of all influenza A-positive respiratory samples collected within our hospital system, regardless of patient exposure history. Between August 2024 and April 2025, we subtyped 4,488 influenza A-positive samples and identified two cases positive for H5 RNA in Alameda County, California, USA. The first case was a 14-month-old girl with mild respiratory symptoms and no H5N1 exposure risks; sequencing of the sample revealed an H5 gene closely related to clade 2.3.4.4b, genotype B3.13 viruses circulating in U.S. dairies. The second case was a 79-year-old male, also with no known exposures, whose sample reproducibly tested positive with a high cycle threshold value but could not be confirmed by public health laboratories. Both patients had evidence of co-infection with other common respiratory viruses. These findings, while requiring cautious interpretation due to low virus levels and the presence of potential confounding factors, highlight limitations in exposure-based testing and demonstrate the potential for cryptic H5N1 circulation. This report underscores that broader, geographically targeted surveillance may be a critical tool for early detection of potential community transmission of pandemic-capable pathogens.