Adobe’s Commitment to Child Safety

Adobe is deeply committed to keeping children safe online and doing our part to fight the spread of child sexual abuse material (CSAM). We have a zero-tolerance policy against any material uploaded to our servers that sexualizes, sexually exploits, or endangers children.

There are several ways we carry out our commitment to keeping children and our users safe:

• We utilize multiple methods to detect CSAM, such as sophisticated machine learning models and hash matching technology. For example, Microsoft’s PhotoDNA and Google’s CSAI Match enable us to compare digital signatures (or “hashes”) of images and videos uploaded by any Adobe user to our servers against databases of known CSAM hashes. We also use Thorn’s machine learning classifier to detect novel CSAM.
• The Trust & Safety team reviews, reports, and removes CSAM discovered by our machine learning models and hash matching technology, as well as CSAM discovered through user reports and account investigations. In all cases, our Trust & Safety team conducts human review of all matches before reporting confirmed CSAM to the National Center for Missing and Exploited Children (NCMEC) as required by US federal law. We review all material reported to NCMEC to ensure a high level of accuracy and quality of reports. We also continue to advance Adobe’s CSAM detection capabilities and build wellness features into our moderation tools to support moderator wellbeing.
• Where possible, we permanently terminate any Adobe user account found to have uploaded or created content sexualizing minors on our platform. In Fiscal Year 2024, Adobe sent 802 CyberTips to NCMEC. We proactively detected 99.9% of the CSAM we reported to NCMEC through machine learning models, hash matching technology, or account investigations.
• As Adobe has transitioned into the new world of generative AI (GenAI), we have maintained our commitment to child safety through safe and responsible GenAI model development. Adobe Firefly is trained with licensed and public domain content where copyright has expired, and we have implemented technical measures to ensure the safety of outputs. Together with the Ethical Innovation team, Trust & Safety facilitates the responsible and safe development of our models through the incorporation of our online safety practices and CSAM mitigation policies into the model development process.
• We implement additional mitigations in our products, based on functionality, to reduce the occurrence of CSAM-seeking behaviors. We use prompt blocking in our GenAI products to prevent users from attempting to generate CSAM. We also incorporate search protections in certain products that present deterrence and intervention messaging when a CSAM-seeking query is identified. This messaging warns the user that CSAM is illegal, provides information on how to report this material to Adobe, and links to mental health resources for users seeking CSAM. Our goal is to interrupt individuals early, before they proceed down an offending pathway, with the goal of preventing further CSAM-seeking behavior and physical child sexual abuse.
• We collaborate with others in industry to fight CSAM at scale and have a longstanding partnership with NCMEC.  We donate technology and technical expertise to NCMEC to assist in their mission to help find missing children, reduce child sexual exploitation, and prevent child victimization. You can read more about Adobe’s partnership with NCMEC here.
• We also support and participate in the Tech Coalition, an organization comprised of technology industry leaders with the goal of preventing and eradicating online child sexual exploitation and abuse. You can read more about the Tech Coalition here.

What you can do

If you encounter material you believe to be child sexual abuse material while using Adobe's products and services, please report it through our in-product abuse reporting mechanisms. You can also make a report directly to the National Center for Missing and Exploited Children through www.CyberTipline.org or by calling 1-800-843-5678.