SF City Agencies Push to Fix Duplicate Image Problem Plaguing Public Records and Housing Databases This Week

San Francisco's Department of Building Inspection and the Planning Department are both contending this week with a wave of duplicate and misattributed images embedded in their public-facing permit and property databases — a technical problem that has stalled document review workflows and complicated the city's already labored effort to speed up housing approvals. The issue surfaced publicly when contractors filing applications through the city's Permit Center at 49 South Van Ness Avenue began flagging rejected submissions tied to images that had been auto-populated with photos from the wrong parcels entirely.

The timing is rough. San Francisco is under a state-mandated housing element that requires the city to plan for roughly 82,000 new units by 2031 — a target the city has openly struggled to meet given entitlement delays, cost pressures, and staffing shortages across planning bureaus. Every day a permit application sits in a queue unresolved because of a database error is a day a project doesn't move.

What Happened This Week

The duplicate image problem appears to have been triggered by a backend migration earlier this year at the San Francisco Department of Technology, which manages shared infrastructure for multiple city agencies. When property record images were transferred between legacy and updated systems, a matching algorithm incorrectly paired some parcel photos with neighboring or similarly numbered addresses. The worst affected areas include dense blocks in the Tenderloin, SoMa, and portions of the Western Addition where parcel boundaries are narrow and address numbering is irregular.

The San Francisco Planning Department's public portal, which residents and developers use to track the status of projects, began showing mismatched exterior photos as early as late June. Contractors working on projects along Market Street and in the Mission District reported confusion during pre-application meetings when staff pulled up images that did not correspond to the sites under discussion. The Permit Center at 49 South Van Ness, which opened in 2020 as a one-stop shop meant to streamline exactly this kind of process, has had to route a subset of applications back through manual review as a workaround.

The San Francisco Budget and Legislative Analyst's Office, in a March 2026 report on permit processing times, found that the average building permit in San Francisco took 138 days from application to issuance — well above peer cities. City officials have pointed to that figure as a key reason housing costs remain high, with median asking rent for a one-bedroom in San Francisco hovering near $2,900 per month as of June 2026, according to aggregated listings data tracked by local real estate monitors. Adding a database error to an already slow system compounds the frustration.

How Agencies Are Responding

The Department of Technology confirmed it is running a systematic audit of image records across affected parcels. The department has prioritized correcting files tied to active permit applications first, then will move to historical records. The Planning Department has instructed its counter staff at the Civic Center offices to manually verify photos for any application flagged with a known mismatched parcel ID — a list that, as of Thursday, included several hundred individual records.

The Department of Building Inspection, which operates separately from Planning and manages its own image repository for code enforcement cases, said its team is cross-checking records against street-view reference imagery to catch mismatches before they cause further delays. Inspectors in the Excelsior and Bayview districts have been asked to confirm parcel photos on-site for a batch of cases currently in review.

For developers and property owners with active filings, the city is recommending they log into the SF Planning public portal and verify that the parcel photo attached to their application matches their address. Applications that appear to have wrong images attached should be flagged using the portal's built-in correction request function, which routes directly to a database administrator. The city expects the bulk of the audit to be complete by July 18, though applications in a manual review queue may see their processing clock paused until the underlying image data is corrected — something applicants will want to factor into their project timelines before the end of the month.