A task new to the Profile: before data is used for training, testing, fine-tuning or alignment of a model, it is analysed for signs of poisoning, bias, homogeneity and tampering and the risks are mitigated as needed. The organisation verifies the data's provenance, where known, and its integrity before use; selects and applies suitable methods for analysing and altering the data, such as anomaly detection, bias detection, cleaning, curation, filtering, sanitisation, fact-checking and noise reduction; and may have a person examine the data through exploratory data analysis.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.