Key Concepts
- 1Consent is not a checkbox but a scope: what data, for what purpose, for how long — and personalization must stay inside the scope the user actually agreed to
- 2Purpose limitation is the core principle: data collected for fulfillment can't be silently repurposed for behavioral targeting without fresh consent
- 3Regulations (GDPR, CCPA/CPRA and successors) grant rights — access, deletion, opt-out of sale/sharing — that the agent's data model must be built to honor, not bolt on later
- 4The agent's memory is personal data: everything it 'remembers' about a user is subject to consent, retention limits, and deletion requests
- 5Sensitive categories (health, precise location, financial) carry heightened requirements; an agent inferring these must treat them with extra care or avoid them
- 6Transparency builds both compliance and trust: showing users what the agent knows and letting them edit/delete it satisfies rights requests and improves the relationship
- 7Data minimization is a feature, not a limitation: collecting only what personalization genuinely needs reduces breach exposure, compliance burden, and creepiness
Why It Matters for PMs
Legal exposure is direct: Consent and privacy violations carry real fines and litigation. The agent's data architecture must honor rights by design, not as an afterthought.
Trust is the currency: Users share more when they trust the boundaries. Transparent, minimal, consent-scoped data use improves both compliance posture and personalization quality.
Retention and deletion are engineering requirements: Because the agent's memory is personal data, deletion and retention limits must be architected in — retrofitting them is costly and risky.
Test Your Understanding
Question 1 (Multiple Choice)
What does 'purpose limitation' require of an agent's data use?
Question 2 (Multiple Choice)
Why is the agent's memory subject to privacy law?
Question 3 (Multiple Choice)
Why is data minimization described as a feature, not a limitation?
Question 4 (Short Answer)
How would you design an agent's data model so deletion and opt-out rights are honored by design?
Think about your answer before revealing the explanation.
Question 5 (Short Answer)
What's the tradeoff between richer personalization data and privacy/consent obligations?
Think about your answer before revealing the explanation.