Web creators' desperate stand: robots.txt pleas and http headers to beg GPTBot not to devour their images—because 'no thanks' totally stops the feast

Artificial intelligence models require vast amounts of training data, often sourced from public websites, prompting concerns among content creators about protecting their original images and work. While it is not possible to completely prevent AI models from training on website images, creators can reduce exposure by clearly stating their boundaries and implementing technical controls. Using a robots.txt file and X-Robots-Tag headers can signal to compliant crawlers that they do not consent to their content being used for training. Additionally, server-level blocking and content delivery networks can enforce these boundaries. Companies like OpenAI, Google, and Anthropic have publicly stated that they honor these headers. WordPress users can utilize plugins like ShortPixel Image Optimizer to restrict AI training on their images. By combining these methods, creators can reduce the risk of their content being used without permission, although determined actors can still spoof headers and bypass these controls.