AI Usage at a Glance
Sep 1, 2023
Customer SvcPractice documented: Civitai deployed CivBot, an AI-powered chatbot built on a large language model and trained on Civitai documentation, to handle user support questions directly on the platform website. CivBot autonomously resolved approximately 72% of Level 1 and Level 2 support tickets, reducing the volume directed to human support staff.
Practice DocumentedView practice →Sep 6, 2023
ProductivityPractice documented: Civitai offers a cloud-based LoRA training service that allows all registered users to fine-tune AI image-generation models on their own datasets without needing a personal GPU. Users upload images, optionally use AI-powered auto-captioning, and the platform's servers run the training job.
Practice DocumentedView practice →Sep 1, 2024
ProductivityNew evidence: 5 minute Rapid LoRA Training on Civitai!, Natural Language Auto-Captioning, Promising New Base Model revealed & More!
Evidence AddedView practice →Nov 14, 2024
Creative GenPractice documented: Civitai offers an on-site video generator that allows registered users to create short AI-generated videos from text prompts or static images, using third-party models such as Kling, Hailuo, Hunyuan, Wan, and Google's Veo 3 integrated into the platform's generator interface. Video generation capabilities expanded significantly through 2024 and 2025.
Practice DocumentedView practice →Jan 1, 2025
ModerationPractice documented: Civitai uses a multi-layered AI system to automatically scan every uploaded image for inappropriate content, combining Amazon Rekognition, a third-party classifier called Clavata, vision-language model scanning, and OpenAI's prompt moderation service. Flagged images are either blocked or routed to human moderators for review.
Practice DocumentedView practice →Jan 22, 2025
RecommendationPractice documented: Civitai offers CivBot, which uses a large language model integrated with Civitai's model database to recommend specific AI model checkpoints and LoRAs to users based on the creative theme or style they describe. This feature is available to all registered users on the platform website.
Practice DocumentedView practice →Mar 14, 2025
Creative GenPractice documented: Civitai offers a web-based AI image generator that lets any registered user turn text prompts into images using thousands of community-uploaded models, including Stable Diffusion, SDXL, and Flux, without needing a personal graphics card. The generator launched in September 2023 and has grown to support advanced workflows like face enhancement and high-resolution upscaling.
Practice DocumentedView practice →Nov 30, 2025
ModerationNew evidence: From Regex to Real-Time: Our 2025 Moderation Journey
Evidence AddedView practice →Civitai offers a cloud-based LoRA training service that allows all registered users to fine-tune AI image-generation models on their own datasets without needing a personal GPU. Users upload images, optionally use AI-powered auto-captioning, and the platform's servers run the training job.
The on-site LoRA Trainer is a web wizard that guides users through uploading a dataset of images, captioning them (manually or via an AI auto-captioning feature powered by the JoyCaption model), and setting training parameters. Civitai's cloud infrastructure runs the training process, producing a LoRA file the user can then download or use directly in the on-site generator. Supported base models include SD 1.5, SDXL, and Flux. Training costs Buzz, with pricing scaled by model type and training complexity. The trainer is open to all registered users. A natural language auto-captioning feature, which automatically describes dataset images using a vision-language model, was added in September 2024.
Civitai offers a web-based AI image generator that lets any registered user turn text prompts into images using thousands of community-uploaded models, including Stable Diffusion, SDXL, and Flux, without needing a personal graphics card. The generator launched in September 2023 and has grown to support advanced workflows like face enhancement and high-resolution upscaling.
Civitai offers an on-site video generator that allows registered users to create short AI-generated videos from text prompts or static images, using third-party models such as Kling, Hailuo, Hunyuan, Wan, and Google's Veo 3 integrated into the platform's generator interface. Video generation capabilities expanded significantly through 2024 and 2025.
Have evidence about Civitai's AI practices? Submit a report.
Submit a report →AI Trace is free and nonprofit. Support our work
Civitai offers a web-based AI image generator that lets any registered user turn text prompts into images using thousands of community-uploaded models, including Stable Diffusion, SDXL, and Flux, without needing a personal graphics card. The generator launched in September 2023 and has grown to support advanced workflows like face enhancement and high-resolution upscaling.
The on-site Image Generator is a cloud-hosted interface for Stable Diffusion and related diffusion models. Users input text prompts and select from tens of thousands of models, LoRAs, and embeddings hosted on the platform; the platform's servers handle all compute. Additional workflows include 'Face fix' (which detects and enhances human faces using a specialized AI pass) and 'Hi-res fix' (which increases image resolution during generation). Access is available to all registered users and costs Buzz, the platform's virtual currency, with a small daily free allowance.
Civitai uses a multi-layered AI system to automatically scan every uploaded image for inappropriate content, combining Amazon Rekognition, a third-party classifier called Clavata, vision-language model scanning, and OpenAI's prompt moderation service. Flagged images are either blocked or routed to human moderators for review.
Civitai's content moderation pipeline has evolved significantly through 2025. Images uploaded to the platform are scanned using Amazon Rekognition and an open-source tagging system to assign content labels and screen for prohibited material. In April 2025, Civitai announced a partnership with Clavata, a policy-driven content classification service, to replace earlier inconsistent systems. In March 2025, Civitai added VLM-based (vision-language model) scanning for images generated on-site. For text prompt moderation, OpenAI's prompt moderation API had been used to scan inputs. Certain tag combinations or resource types trigger a mandatory human moderator review before content becomes publicly visible. The system is combined with community-driven 'Knights of New Order' rating, in which users help classify borderline content. In 2024, the platform processed over 586,800 reports and removed over 730,000 images according to the company's own transparency report.
