New York Requires Disclosure of AI-Altered Rental Photos

⏱️ Lectura: 10 min

As of this week, posting a rental listing in New York with AI photos without disclosing it can count as a deceptive practice the city can prosecute. Mayor Zohran Mamdani presented the Rental Ripoff Report on July 16, 2026, which recommends requiring landlords and real estate agencies to disclose when a listing image was generated or altered with artificial intelligence.

📑 En este artículo

1. TL;DR
2. What happened
3. Context and background
4. Technical details: how to detect AI photos in a listing
5. How to try it
6. Impact and analysis
7. What’s next
8. Frequently Asked Questions
   1. Is this rule already law in New York?
   2. What happens if a landlord uses AI just to clean up clutter in a room?
   3. Does the C2PA standard work with any camera or editor?
   4. Does this mechanism work for rental platforms in Latin America?
   5. What other tech measure did Mamdani announce that same week?
   6. Who backed the Rental Ripoff Report besides Mamdani?
9. References

The measure arrives in a context where AI photos already distort how thousands of apartments for rent actually look, a problem that, according to the report itself, goes far beyond New York’s five boroughs.

TL;DR

• Mayor Zohran Mamdani published the Rental Ripoff Report on July 16, 2026.
• The proposal requires landlords and real estate agencies to disclose when a rental photo was altered or generated with AI.
• The report grew out of the Rental Ripoff Hearings, where Mamdani’s team heard from 2,400 New Yorkers across the five boroughs.
• Deputy Mayor Leila Bozorg and Director Cea Weaver backed the report along with the recommendation to recognize tenant unions.
• The C2PA standard (Content Credentials), backed by Adobe, already allows signing the origin of an AI-edited image.
• The open source tool c2patool audits an image file’s provenance metadata from the command line.
• The report comes one day after Mamdani announced a click-to-cancel rule aimed at companies like Adobe.

What happened

The Rental Ripoff Report is the result of the so-called Rental Ripoff Hearings, hearings Mamdani opened during his first week in office. His office met with 2,400 New Yorkers in each of the five boroughs to hear complaints about housing conditions.

Among the report’s recommendations is requiring rental listings to disclose when they were altered with artificial intelligence or other digital tools, without distinguishing between generating an image from scratch or simply retouching a real photo.

“At the hearings, we heard from thousands of New Yorkers living with mold that was never treated, pests that were never addressed, and charges that were never explained,” Mamdani said when presenting the report. Deputy Mayor for Housing and Planning, Leila Bozorg, added that the policies “are rooted in real experiences.” Cea Weaver, director of the Office to Protect Tenants, described the report as part of “a new chapter in tenant power” in the city.

The package of recommendations isn’t limited to AI photos: it also includes formally recognizing tenant unions and modernizing the city’s code enforcement systems.

Context and background

Misleading images in rental listings aren’t a problem exclusive to New York. PetaPixel’s own report describes them as “an increasingly serious problem,” visible on listing platforms worldwide long before generative AI made photo retouching mainstream.

The situation becomes more serious when a tenant has to sign a lease remotely, for example when moving for a new job, without being able to visit the property before committing to a deposit.

This wasn’t the only tech-related measure of the week. One day earlier, on July 15, 2026, Mamdani announced a click-to-cancel rule aimed at software companies like Adobe, in the same vein of demanding transparency and consumer control over digital tools.

The editorial pattern is consistent: regulate not the technology itself, but the lack of disclosure when it’s used to influence a resident’s financial decision, whether signing up for a subscription or signing a lease.

Technical details: how to detect AI photos in a listing

The report doesn’t specify a technical verification mechanism, it only mandates disclosure. That leaves the door open for rental platforms to adopt different approaches to detect or audit AI photos in their catalogs.

The most robust one is C2PA (Coalition for Content Provenance and Authenticity), the standard behind Content Credentials that Adobe integrates into Photoshop and its generative fill feature. When a C2PA-compatible editor saves an image, it embeds a cryptographically signed manifest listing the actions applied, including AI generation.

Method	When to use it	Advantage	Limitation
Content Credentials (C2PA)	When the editor (Photoshop, Firefly) supports the standard	Cryptographic, verifiable proof of origin	Lost if the image is recompressed or screenshotted
EXIF heuristic (Make/Model/GPS)	Quick filter at photo upload time	Requires no external dependencies or network access	Legitimate WhatsApp photos or screenshots also lose EXIF data: many false positives
Visual classifier (CLIP-type)	Post-hoc audit across the whole catalog	Detects manipulation even without metadata	Requires its own training and isn’t 100% accurate, needs human review

💡 Tip: if your rental platform already integrates Adobe Firefly or Photoshop into its photo upload flow, checking the C2PA manifest is more reliable than building your own classifier.

A minimal example to check whether a listing photo still has camera metadata, a weak but quick signal that it didn’t go through an image generator:

import exifr from "exifr";

async function tieneMetadataDeCamara(rutaImagen) {
  const datos = await exifr.parse(rutaImagen, ["Make", "Model", "GPSLatitude"]);
  const sinCamara = !datos || (!datos.Make && !datos.Model);
  return { sospechosa: sinCamara, datos };
}

const resultado = await tieneMetadataDeCamara("depto-recoleta-living.jpg");
console.log(resultado.sospechosa ? "Check: no camera metadata found" : "Camera metadata present");

This check uses the exifr package to read the Make, Model, and GPSLatitude tags. If none is present, the photo gets flagged as suspicious for manual review, although the result alone proves nothing.

The full flow for a platform looking to comply with New York’s disclosure requirement would look like this:

flowchart TD
 A["Landlord uploads photo"] --> B["Editor applies AI (optional)"]
 B --> C["C2PA manifest is generated"]
 C --> D["Platform verifies with c2patool"]
 D --> E["Listing shows 'Altered with AI' label"]
 D --> F["Listing is published without a label"]

How to try it

To check whether an image carries a C2PA manifest, the go-to tool is c2patool, maintained as open source by the Content Authenticity Initiative. It’s installed with Rust and Cargo, available on Windows, macOS, and Linux:

# macOS / Linux (with Rust installed)
cargo install c2patool

# Windows (PowerShell, with Rust installed)
cargo install c2patool

With the binary installed, dumping the manifest of a listing photo is a single command:

c2patool depto-recoleta-living.jpg -d

If the image has Content Credentials, the output is a JSON object with the manifests key populated, including the actions applied (editing, AI generation, cropping). If that key comes back empty or the command fails, the image doesn’t carry a manifest, which doesn’t confirm it’s real, only that it doesn’t disclose its origin.

A more realistic example integrates this check into a rental platform’s upload pipeline, running c2patool as a subprocess and storing the result per property:

import { execFile } from "node:child_process";
import { promisify } from "node:util";
const run = promisify(execFile);

async function auditarFotoDeListado(propertyId, rutaImagen) {
  let manifiesto = null;
  try {
    const { stdout } = await run("c2patool", [rutaImagen, "-d"]);
    manifiesto = JSON.parse(stdout);
  } catch {
    manifiesto = null;
  }

  const generadaConIA = manifiesto?.manifests
    ? Object.values(manifiesto.manifests).some(m =>
        m.assertions?.some(a =>
          a.label === "c2pa.actions" &&
          a.data.actions?.some(act => act.softwareAgent?.includes("AI"))
        )
      )
    : null;

  return { propertyId, generadaConIA, tieneManifiesto: Boolean(manifiesto) };
}

The function returns an audit record per property that a platform could use to decide whether to show the disclosure label required by Mamdani’s report.

⚠️ Watch out: c2patool only detects manifests that the editor actually embedded. A screenshot, a WhatsApp recompression, or an editor without C2PA support erase that proof without erasing the manipulation.

Impact and analysis

For those building proptech in Latin America (platforms like Properati, Inmuebles24, or Mercado Libre’s real estate section), this rule doesn’t apply yet, but it marks a clear direction: rental platforms will need to disclose the origin of their photos before a local law requires it.

The technical cost of getting ahead of this is low. Adding a C2PA manifest check to the photo upload pipeline, like the one in the example above, doesn’t require training any custom model, just running c2patool against each file before publishing it.

The honest limitation is that disclosure depends on the image editor having embedded the manifest. Image generation tools that don’t adopt C2PA, or a simple crop with an app lacking that support, leave the photo without any verifiable trace. In practice, enforcement of the rule in New York will depend more on tenant complaints than on foolproof automatic detection.

That also explains why the report doesn’t mandate a specific technology: requiring disclosure is easier to legislate and enforce than requiring a technical standard the whole industry hasn’t adopted yet.

What’s next

The Rental Ripoff Report is, for now, a set of recommendations from the mayor’s office, not a law. For the AI photo disclosure requirement to become enforceable, it needs to go through the New York City Council, the same path the formal recognition of tenant unions included in the same report will have to follow.

The click-to-cancel rule announced the day before, aimed at companies like Adobe, follows a separate regulatory process, which suggests the Mamdani administration is building a broader tech transparency package in 2026, not an isolated measure.

📖 Summary on Telegram: View summary

Try it yourself: run c2patool tu-foto.jpg -d on the last image you uploaded to a rental listing site and check whether it carries an origin manifest.

Frequently Asked Questions

Is this rule already law in New York?

Not yet. The Rental Ripoff Report is a set of recommendations from Mamdani’s office; to become mandatory, it needs to be approved by the New York City Council.

What happens if a landlord uses AI just to clean up clutter in a room?

According to the report, any alteration of the listing with AI or digital tools should be disclosed, regardless of the type of retouching applied.

Does the C2PA standard work with any camera or editor?

No. Only cameras and editors that integrate Content Credentials, such as some recent camera models and software like Adobe Photoshop, embed the manifest natively.

Does this mechanism work for rental platforms in Latin America?

The technical mechanism, C2PA plus metadata auditing, is replicable in any proptech company, although no city in the region currently requires AI photo disclosure by law.

What other tech measure did Mamdani announce that same week?

One day earlier, on July 15, 2026, he announced a click-to-cancel rule aimed at software companies like Adobe.

Who backed the Rental Ripoff Report besides Mamdani?

Deputy Mayor for Housing and Planning Leila Bozorg, and Cea Weaver, director of the Office to Protect Tenants.

References

• PetaPixel: original coverage of the Rental Ripoff Report and the statements from Mamdani, Bozorg, and Weaver.
• NYC.gov: the official website of the City of New York where mayoral office reports are published.
• contentauth/c2patool on GitHub: open source tool for auditing C2PA manifests in images.
• MDN: File API: reference for reading files and image metadata in the browser or Node.js.
• Wikipedia: Synthetic media: general background on AI-generated or AI-altered content.

📱 Enjoying this content? Follow @programacion on Telegram for daily tech content in Spanish: quick summaries, fresh content every day.

Imagen destacada: Foto de Kiran Naidu en Unsplash