Marching Toward Dystopia - Reverse Image Databases
A few weeks ago, while I was riding public transportation, a fellow rider began to behave erratically toward me. He looked like any other professional on the train; he was well kept and carrying a corporate bag. He followed me onto the platform, stood two inches from me while I used the machine to pay for my next ride, and would not leave me even after I told him numerous times that he was too close and making me uncomfortable.
I ran to my next connection and shed him without thinking twice. The odds of seeing him again were slim to zero. Now imagine an alternate universe where this man can press a button on a pair of glasses and gain access to my full name, my address, and my place of employment.
Last year, two Harvard students demonstrated how easy this would be using a modified version of Meta's smart Ray-Ban glasses combined with the reverse image capabilities of PimEyes and the data broker ecosystem. Facial recognition is not part of Meta's current smart glasses product, but as of May 2025, Meta did indicate that this feature would be in the next release.
The truth is that someone does not need to have access directly from their glasses. As it stands, it is shockingly easy for someone to snap a photo of a stranger, run their face through a consumer reverse image tool, and use the results to infer their identity. With a name and approximate location, one could then leverage the massive data broker ecosystem to find their home address, phone, known relatives, approximate net worth, employer, and other alarming personal data primarily gleaned and aggregated from public data sources. The harms of this practice are already well documented, and I encourage you to read more and consider removing yourself from these databases either manually or by paying for a subscription.
On a scale of 1-10 for technical skill, I would rate this process as requiring a knowledge level of about a 5. The glasses reduce this level to a 1 - it would be near effortless.
The author and journalist Kashmir Hill has written extensively about the erosion of privacy and physical safety from facial recognition technology. Her book, Your Face Belongs to Us, primarily focuses on Clearview.ai, one of the earliest mainstream adopters of this technology. In May of 2022, a judge ruled that Clearview would be permanently banned from offering its technology to most private corporations and public entities, restricting its use primarily to law enforcement. What has occurred over the last 1-2 years is that the consumer version of this technology has taken a leap forward due to sophisticated AI models that can now crawl through large swaths of data and catalog it for individual use.
The growth of reverse image databases has been exponential even in the last 12 months, and several are picking up where PimEyes left off by indexing public social media posts.
Imagine every time a person sitting near you snaps a photo of themselves, inadvertently capturing you in the background, and sharing this to TikTok, Instagram or any social media platform. What could a possible bad actor do with multiple data points that contain your face? They could have a very strong idea where you live, work, and your typical daily patterns.
We found a tiny shot of my spouse in the far background of someone else's recreational photo very close to our home. We also found that his image (along with our child's) was used by a stock photo provider from an event where we likely unknowingly provided consent just by purchasing a ticket.
Some of my false matches on several sites appeared to be from pornography sites, which opens up numerous other concerns about what happens if a prospective employer or personal contact attempts to validate your identity using blatantly false information. There have been known incidences of facial recognition incorrectly identifying individuals and leading to arrests. The consumer version of this could lead to harassment, ostracization, and/or loss of employment.
Individuals lack sufficient recourse in these scenarios, other than relying on someone's good graces to remove the photo (and even then, the reverse image provider keeps a cached copy), or to use the sketchy opt-out form for these providers, which often requires you to send (redacted) identification to prove ownership of your face.
Below is a list of the sites I have personally tested and successfully opted out of at the time of this writing:
What can you do?
While no single action will guarantee anonymity, there are steps we can all take to reduce our digital footprint.
Search for your image on these sites to facilitate its removal from public platforms you manage
Contact the owner of any websites or public social media post where your image may have been used without your consent and request that they remove it
Leverage the opt-out forms for these and other sites as they appear to remove your data from their search
Run a check at least once per quarter to see if new results appear and handle as needed
Remove your information from the data broker ecosystem
Lobby your federal and state representatives for privacy legislation that would make this practice illegal
Engage with your network (colleagues, friends, relatives, etc) to help raise awareness of the dangers of this technology and the steps they can take to protect themselves and their loved ones
Support critical institutions such as Ludlow Institute, whose mission is to raise awareness about the importance of privacy