For several years, the creation and distribution of pornographic deepfakes—images or videos that use artificial intelligence to manipulate real people—have exploded on the Internet. Three researchers, Sarah Morgan, Sophie J. Nightingale, and Hany Farid, have pointed to the responsibility of major service providers such as Cloudflare, Namecheap, and Google in the distribution of these often violent images. These images can cause significant harm to victims-survivors, pushing them into social rupture. In May 2025, MrDeepfakes, one of the most popular sites for producing this type of content, closed, stating that "a key service provider has definitively terminated its services. The data loss has made it impossible to continue the activity."
In a study published in the Journal of Online Trust and Safety, the researchers found that a few of the largest global Internet companies provide the basic infrastructure—such as web hosting and content delivery networks—for deepfake sites. They cited Cloudflare, Namecheap, and Google as key players. The number of deepfakes detected has increased from 500,000 images in 2023 to 8 million in 2025. While these figures are estimates of deepfakes in general, a separate study indicates that most are created with the intent of targeting women to produce sexualized images without their consent.
The researchers examined "the role that the general technological infrastructure—often invisible—plays in the creation and distribution" of these deepfakes by analyzing 88 sites hosting celebrity deepfakes found through keyword searches and Google alerts. Of these, 38 were exclusively dedicated to sexist and sexual deepfakes. Their analysis found that most (80) were adult video sites, five were adult forums. The remaining three included a general forum with a specific discussion thread on the subject, a subscription-based creator platform with adult content, and a cloud-based file-sharing site.
The researchers identified five dominant infrastructure providers: Cloudflare, Google, Namecheap, WordPress, and Protonmail. Cloudflare provided the majority of the services, ranking first in four of the nine infrastructure components: website hosting, content delivery network (CDN), domain name system (DNS), and analytics tools. Google had a dominant position in the provision of SSL certificates and advertising spaces. Namecheap, WordPress, and Protonmail were the main providers of domain name registration, content management systems, and messaging services. The findings were consistent across the subcategory of sites dedicated exclusively to this type of content.
The researchers argue that the proliferation of these sexist and sexual deepfakes is not just a problem of content creation or moderation but also an infrastructure issue. They suggest that all infrastructure providers should permanently block sites that host exclusively or mainly content related to this type of material. However, such bans should be reserved for the most egregious cases and not taken lightly. They reference Cloudflare’s 2017 decision to close the account of the neo-Nazi site The Daily Stormer, which showed that infrastructure providers can implement such decisions.
Since then, Cloudflare has been cautious about such actions, as seen in its stance toward 8chan after the El Paso and Dayton shootings in 2019. The company stated that it would no longer allow 8chan to use its services, citing the platform's role in inspiring tragic events. Similarly, Cloudflare blocked Kiwi Farms, a forum known for organizing harassment campaigns, despite the difficulty of such decisions. However, even when sites are blocked, malicious users can often recreate accounts using new addresses. The challenge lies not just in blocking but in holding individuals accountable, a task complicated by the anonymity of many actors and the lack of consistent legal frameworks across jurisdictions.
Major Internet Service Providers Face Scrutiny Over Role in Deepfake Distribution
AI-rewritten from original reportingHow it works
deepfakestech-infrastructureonline-safetyinternet-giantsdigital-abusecontent-moderation
Original sources:
- 🇫🇷Next



