FAQs

  • The DTOM Declaration is a watermark that can be embedded into your digital assets that signifies to AI systems and companies that the it should not be used for AI training without proper consent.

  • Currently, the DTOM Declaration can be used on image files only, including digital art and photographs. Multiple image formats are supported (JPEG, PNG, WEBP, TIFF).

    We are currently working towards adding robust support for videos. We also plan to support audio and text.

  • DTOM takes a privacy-first approach to handling your creative works and personal information. When you use our service, we maintain a strict data minimization policy:

    For Images:

    • We only process your image temporarily to create the DTOM watermark declaration

    • Once you download your watermarked file, your original image and the processed version are immediately deleted from our systems

    • We do not maintain any image database or archive

    • We do not store, cache, or retain any copies of your images after processing

    For Personal Information:

    • The only personal information we collect is email addresses, and this is strictly optional

    • Email addresses are only collected if you explicitly opt in to our newsletter for updates about the DTOM initiative

    • We do not collect any other personal data such as names, locations, device information, or usage patterns

    • We do not track which images are processed or maintain records of watermark creation

    This approach ensures that you maintain complete control over your creative works while using our service. The temporary processing is solely for the purpose of applying the DTOM declaration watermark, and our systems are designed to automatically purge all image data once you have downloaded your watermarked file. This aligns with our core mission of empowering creators to protect their rights while respecting their privacy and data sovereignty.

  • While robots.txt is anchored to websites, the DTOM Declaration is is anchored to content. This means that you, as the creator of the content, can make the decision to opt-out.

    In order for this to work well, the DTOM Declaration must stick with the content wherever it gets published. Metadata is easily lost, for instance when exporting from an iPhone or when uploading onto Instagram. Watermarks will not be lost, and any intentional removal violates DMCA 1202(b).

  • The watermark is invisible: it is designed to be only machine-readable, and should not impact human perception.

    Artifacts may be visible on certain types of images, such as very small images (e.g. 128-by-128 thumbnail) or very “flat” images (e.g. a screenshot of text). The DTOM team is working on further optimizations.

  • We provide a free tool that allows you to add (and check for) a DTOM Declaration. All you need to do is navigate to the app and drag & drop your files.

  • Currently, all DTOM Declarations must be added individually through the web app. We are working on a Photoshop integration (among others), to launch in 2025.

  • With enough effort, it can, but doing so intentionally would violate DMCA 1202(b). The DTOM Declaration is designed to be robust against common “unintentional” image modifications, such as resizing or cropping.

  • While it may not be legally binding yet, it is a compelling step towards a standard that pushes for AI compliance and sets the groundwork for broader legal recognition.

  • By making non-compliance a public issue, we incentivize companies to respect creators’ rights. Currently, DTOM is young, so while we have a clear vision for a long-term solution, much more work needs to be done. By lending us your voice, you can help push this endeavor forward.

  • Yes, we provide free public tools so that you can add and check for DTOM Declarations, and will ensure that it remains so.

  • Most existing opt-out solutions are applied at the website-level, which means that websites can say “don’t train on the stuff we post.” Apple’s opt-out in particular uses robots.txt.

    DTOM is applied at the content-level, which means that creators can say “don’t train on me,” regardless of which website it gets posted on. This gives creators much more control.

  • Glaze makes it difficult for AI models to train on content.

    ArtShield applies a watermark that Stable Diffusion (once) used to mark their AI-generated content.

    DTOM is like a robots.txt but for content rather than websites.

  • The DTOM Declaration is compatible with most tools. For Glaze in particular, we recommend applying Glaze first before adding on the DTOM Declaration. If there is a use case you would like specific clarification on, email us.

  • This functionality is not manageable within the scope of DTOM. This is largely because DTOM, by nature, is anonymous: no unique data is stored.

    However, we are planning to partner with Trufo to provide these services, coming later in 2025. This should allow you to prove ownership, prove authenticity, and monetize IP.

  • Shoot us an email, or send a message on X, Instagram, or Discord! We will do our best to continue making improvements.

  • Yes, and our team has a lot of great features & functionalities planned. The more feedback we get from the community, the better the tools we can build.

    Sign up for our newsletter to stay tuned.

  • The organization committee consists of lawyers, engineers, and artists, united by the mission to ensure that human creativity is respected in the age of generative AI.

  • Email us with your résumé and a brief statement of intent. We will get in touch.

  • You can support the DTOM initiative by joining the movement, adding DTOM Declarations, and spreading the word to other creators.

    We currently do not take donations. We are planning to file for 501c3 status.