Data security protects data from unauthorized access, disclosure, alteration, loss, or destruction. In an ML project, security requirements belong in pre-project planning alongside data access, privacy, and the architecture that will support delivery. Missing these requirements can expose the organization to liability or increase costs after plans are set. section 2.0 section 2.8
Turn security concerns into project decisions:
- Identify the data assets, how the team will access them, and what capability it has to prepare them for ML. section 2.8
- Document security and privacy requirements before estimating the resources needed for delivery. section 2.0 section 2.8
- Check applicable data-protection and domain laws. Record corporate-responsibility and ethical requirements as project assumptions, and create privacy and algorithmic impact assessments. section 2.6
Unknown personal-data origins
If the origin of personal data is unknown, its use may violate consumers’ privacy and lack the data owners’ consent. Resolve sourcing and usage constraints before relying on that data for an ML project. section 2.0
