“To utilize top-tier AI models, ordinary users become privacy-conscious geeks.”

“To use top-tier AI models, ordinary users become privacy-conscious geeks.”

Regular users research device fingerprints and anti-fingerprinting tools to use better AI models, causing privacy concerns to spread from the advertising sector to generative AI. Anthropic banned some Chinese users due to the Prompt Steganography mechanism. OpenAI users reported that the model became less effective and tried to modify it by changing time zones and using fingerprint browsers to bypass risk controls. Currently, websites generate device identifiers through software and hardware characteristics (such as Canvas, WebGL, CPU parameters) for login verification and fraud prevention. However, e-commerce platforms like Amazon also widely collect such information for security control or cross-site tracking. Although device fingerprints can identify abnormal attacks, using them for user profiling or advertising attribution involves privacy risks. Under GDPR and ePrivacy regulations, if this technology is used to distinguish specific users or create long-term identifiers, it may constitute online identifiers or personal data, and the principles of necessity, purpose limitation, and minimum collection must be followed.