Artificial intelligence (AI) increasingly performs tasks that previously only humans do. The intention is that machine learning (ML) will automate processes that rely on image recognition and free up humans for higher-value tasks. We have identified many lab reports used in Australia and overseas for water damage and mould assessments that exploit “state of the art equipment that is of the most recent technology available for fungal spore identification and quantification”. However, many of these data sets suggest serious concerns about accuracy, where spore counts sometimes span a range over 5-orders of magnitude and appear numerically implausible for some traps. This can have serious ramifications for air quality interpretations, for risk assessment of exposure, or as part of legal cases. Standard D7391-20 covers the method of categorization and quantification of airborne fungal structures where a human analyst is the detector extracting the ground truth on slides by manually labelling and comparing against reference texts and microscopic mounts from known sources. Negative and positive bias occurs when spores are obscured, the AI mistakes spores for non-fungal particles or other elements are mistaken for spores. Unfortunately, the best AI’s are the least transparent since deep neural networks entail the black box effect, where internal bias may lead to incorrect classifications or embedded errors become amplified over time. Assessments using AI/ML cannot be relied on where interpretations remain unqualified by additional, non-AI, mould assessment methods. Where assessments use AI/ML, this must be disclosed to the client by the occupational hygienist or anyone acting in that role including insurance assessors. Each stakeholder must be informed that data analysis has been performed by AI/ML and hence is subject to various forms of bias that could invoke various legal claims for liability, negligence and conflict of interest. These risks need to be disclosed otherwise the assessor vitiates a clients informed consent and could violate a more general obligation to warn the client about potentially harmful consequences.