NSFW Image Prompt Verification

I’m building a personal dataset that pairs text prompts with images, many of which are explicit. Your job is straightforward but detail-oriented: open each image in our browser-based tool, read its associated prompt, then decide whether that prompt truly describes what you see.

If it’s a clean, one-to-one match, simply confirm it. When the image and prompt diverge in any way—partial or total mismatch—you’ll flag the case through the same internal system. No additional notes or bounding boxes are needed; the platform handles everything once you select the appropriate option.

Because a significant portion of the content is NSFW, you must be fully comfortable viewing adult material for extended periods and able to maintain objective judgment throughout the review process.

Deliverables:
• Verified “match” or “mismatch” status recorded for every assigned image-prompt pair inside the internal tool.

Consistency, speed, and an eye for subtle differences are what I value most. Let me know your typical hourly throughput and when you can start.