Required control
Minimum data. Defined purpose.
Every field has a documented purpose. Sensitive data gets stricter review and access controls.
Request access
Menu
Privacy, fair housing and responsible AI
Housing is consequential. Our safeguards should be, too.
Rentonomy is building an AI-native operating platform for rental housing. Privacy and fair housing have to shape the data, models, workflows, approvals and evidence from the start.
01
Privacy by design
Collect less. Explain why. Protect it for its full life.
Minimize what enters the system, make every use understandable, and scale controls to how sensitive the data is.
Required control
Clear resident rights.
Residents can learn what we hold and request access, correction, portability or deletion where applicable.
Required control
Access follows the job.
Least-privilege roles, strong authentication, scoped service access, encryption and reviewable access logs.
Required control
Retention is a schedule, not a habit.
Retention tied to legal, operational and safety needs, with deletion from active systems and consistent handling in backups.
Required control
Models and vendors stay inside boundaries.
Subprocessors need a defined purpose, contractual limits, security review and deletion obligations.
Required control
Incidents meet a practiced response.
A written plan from detection through lessons learned, exercised before it’s needed.
Your data stays yours
Two bright lines.
We never sell resident or customer personal information.
We never use customer or resident content to train shared models without explicit authorization.
02
Fair housing by design
AI should make housing decisions more accountable. Never less.
The federal Fair Housing Act protects against discrimination based on race, color, national origin, religion, sex, familial status and disability. State and local law may add more. We commit to the strongest applicable standard in each market.
Required safeguard
Neutral criteria, applied consistently.
Written, versioned, legally reviewed policies. Penny does not invent new standards for individuals or properties.
Required safeguard
No protected traits or proxy shortcuts.
Protected traits can’t drive decisions. Proxy inputs need review, documented necessity, testing, and removal when the risk isn’t justified.
Required safeguard
Humans own consequential decisions.
AI can gather facts, organize policy, draft communications and recommend next steps. Final consequential decisions need authorized human review.
Required safeguard
Every decision can be explained.
Records show the inputs, policy version, model or workflow, recommendation, approver and final action.
Required safeguard
Test outcomes, not only intentions.
Testing before launch and after every change for accuracy, error rates and materially different group outcomes, with remediation.
Required safeguard
Accommodation and accessibility are first-class.
Accessible request paths, prompt review, only the information needed, no penalty or retaliation, and a clear record.
03
The human gate
Penny can prepare the decision. She cannot own the consequence.
An authorized human must approve before Penny can take any of these actions.
Approve or deny access to housing
Change rent, deposits, fees or individual lease terms
Reject a reasonable accommodation request
Send a final adverse-action or eviction notice
Override operator policy or required legal review
Authoritative baseline
Standards we are using to build the standard.
These primary sources inform our design and governance work. Linking them does not represent agency endorsement, certification or an audit of Rentonomy.