1 min lesson
PII: classify, then control
Walk through "PII: classify, then control" in order, then name the proof that tells you it worked.
Step 1 of 2
PII: classify, then controlyou can't protect what you haven't found
- 1Classify first. Tag columns by sensitivity (PIIPersonally Identifiable Information. Data that can identify a person (names, emails, SSNs); regulated and sensitive. Press Enter for the full definition., code content, internal-only). You can't apply a policy to data you haven't labeled.
- 2Mask or tokenize at rest. Replace raw PIIPersonally Identifiable Information. Data that can identify a person (names, emails, SSNs); regulated and sensitive. Press Enter for the full definition. with masked or tokenized values so most consumers never touch the original.
- 3Column- and row-level security. Grant the marketing analyst the aggregate, not the email; scope rows by tenant where it applies.
- 4Least privilege by default. Access is requested and granted, not the ambient default - and it's reviewed, not granted forever.
"We need this data to make the agent better" is true and is also exactly where privacy discipline slips. Code and usage data is some of the most sensitive a developer owns. Expanding capability means building the guardrails alongside it: opt-in/consent handling, masking of secrets and PIIPersonally Identifiable Information. Data that can identify a person (names, emails, SSNs); regulated and sensitive. Press Enter for the full definition. before training use, isolated environments and an audit trail - not shipping the capability now and promising controls later. Saying this unprompted in the interview signals you understand the stakes of the role's flagship project.