Newsroom
Where we use AI
We use AI as a tool, not a byline. A human is always accountable.
Our principle
AI can speed up drafting, translation, and pattern-finding. It cannot be responsible for a decision that affects a worker's money, safety, or livelihood, and it cannot be the author of record for anything we publish. A named person owns every output.
We use it
Translation and transcription
Interviews conducted in Hindi, Tamil, Bhojpuri, and other languages are transcribed and translated with AI assistance, then checked by a fluent human before use.
We use it
Data pattern-finding
We use AI to surface anomalies in operations data, an unusual spike in deductions, an outlier in occupancy, which our team then investigates manually.
We use it
Draft assistance
Reporters may use AI to structure an early draft or summarize source documents. Every published sentence is written and verified by a human.
We do not
Autonomous publishing
No article, correction, or data release is published without human review and a named editor signing off. AI never publishes on its own.
We do not
Fabricate quotes or images
We never use AI to generate quotes, synthetic people, or images presented as real. Illustrative visuals are labeled as such.
We do not
Decide member outcomes
Decisions about a member's housing, pay, or eligibility are made by people accountable under our policies, never by a model alone.
Accountability
If we make a mistake with AI assistance, it is our mistake, not the tool's. It is corrected under the same editorial standards and logged on our corrections page as any other error would be.
Questions about how we use AI?
We are happy to explain our tooling and safeguards in detail.