Photobucket, the once-popular photo-storage platform from the Myspace era, is planning to license user-uploaded images to AI companies for training their machine learning models. CEO Ted Leonard confirmed in an interview with Business Insider that while no images have been used for AI training yet, all photos marked as “public” will be eligible for licensing deals with AI companies.
The policy creates a controversial situation for users with dormant accounts. To opt out of AI licensing, users must reactivate their accounts and switch settings to private—but reactivation requires a paid subscription starting at $5 per month. While users can delete their accounts for free, doing so prevents them from viewing or downloading their old photos first.
Founded in 2003, Photobucket became a cultural phenomenon during the Myspace era, serving as a primary image-hosting service for social media users. Unlike competitors Flickr and Webshots, Photobucket never deleted old user photos, meaning images uploaded as far back as 2005 remain on their servers. This extensive archive now represents a potential goldmine for AI training data.
Photobucket estimates that approximately half of its 13 billion stored images are set to public and therefore eligible for AI licensing. Leonard described the licensing model as similar to advertising revenue from a decade ago, providing capital to support the company financially and fund product development. He indicated that licensing deals would typically span a few years and target specific content types, with AI companies requesting particular categories like “landscapes with people in the distance” or “beach and mountains.”
The company sent an email notification in July updating its terms of service, which now includes language about “scanning and processing of the Public User Uploaded Content, including extracting physical features, e.g. measurements, of your Biometric Information (e.g., face, iris, etc.), for the purpose of artificial intelligence and machine learning training.”
The challenge for many users is accessibility—some may have signed up with old student email addresses they can no longer access, or simply forgotten their login credentials entirely. Leonard acknowledged the company has been attempting to reach customers through email, on-site alerts, and press coverage, but admitted many users likely ignored the terms of service update emails, as most people typically do.
Key Quotes
I view licensing content specifically for AI training as similar to the advertising revenue the company generated 10 years ago. It is a vehicle for continuing to support the company from a financial perspective, and it gives us capital at what we think will be fairly significant in material margins to continue investing in the product itself.
CEO Ted Leonard explained the business rationale behind licensing user photos to AI companies, positioning it as a modern revenue stream comparable to traditional advertising models that sustained the platform previously.
A company will come to us and say, ‘We need content of landscapes with people in a distance, and we’re looking for beach and mountains.’ So we say: ‘OK, great. Based on that request, we have 150,000 images that we can license you.’
Leonard described how AI licensing deals would work in practice, with companies requesting specific categories of images rather than blanket access to all content, suggesting a more targeted approach to AI training data acquisition.
There’s kind of a two-way street where if you put a whole bunch of personal memories on the site, maybe you should do your best to continue to update the email address and everything like that.
Leonard placed some responsibility on users to maintain their account information, though this statement may strike many as tone-deaf given that users uploaded content under different expectations and terms of service years or decades ago.
Our Take
Photobucket’s AI licensing strategy exemplifies the ethical gray areas emerging as AI companies scramble for training data. While technically legal—public photos can be licensed—the moral dimension is murkier. Many users uploaded content 15-20 years ago under entirely different expectations about how their data would be used. The pay-to-opt-out model is particularly problematic, creating a financial incentive structure that benefits from user abandonment and inaction.
This case also reveals how legacy platforms are finding new life through AI monetization, transforming what might have been considered digital archives into valuable commercial assets. As AI regulation evolves, expect increased scrutiny of these practices, particularly around biometric data extraction mentioned in Photobucket’s updated terms. The broader lesson: your digital past is never truly forgotten, and may be repurposed in ways you never anticipated.
Why This Matters
This development highlights a growing trend of legacy internet platforms monetizing user-generated content through AI licensing deals, raising significant questions about digital consent and data ownership. As AI companies face increasing scrutiny over training data sources and copyright concerns, they’re turning to platforms with vast repositories of user content to build their datasets legally.
The Photobucket case is particularly significant because it demonstrates how dormant digital assets from the early social media era are being repurposed for modern AI applications without active user engagement. The requirement to pay for account reactivation to opt out creates a financial barrier that effectively monetizes user inaction.
This follows similar moves by Meta, which used public Facebook photos for AI training, and other platforms that have incorporated user content into AI datasets. The practice raises broader questions about the long-term implications of terms of service agreements and whether users truly understand how their digital footprints may be utilized years or decades after initial upload. As AI training data becomes increasingly valuable, expect more platforms to follow Photobucket’s lead in monetizing historical user content.
Recommended Reading
For those interested in learning more about artificial intelligence, machine learning, and effective AI communication, here are some excellent resources:
Recommended Reading
Related Stories
- Photobucket is licensing your photos and images to train AI without your consent, and there’s no easy way to opt out
- Tech Tip: How to Spot AI-Generated Deepfake Images
- Jenna Ortega Speaks Out Against Explicit AI-Generated Images of Her
- The Artificial Intelligence Race: Rivalry Bathing the World in Data