Interest inference is the process of estimating a person’s preferences from observed behavior rather than from an explicit statement. In digital systems, signals such as searches, clicks, viewing time, replays, or skipped content can be used to predict topics a user may find relevant.
In cyber security and privacy, it matters because small or accidental actions can contribute to a profile and influence ad targeting or recommendations. A brief pause on a video, for example, may be treated as a weak signal of interest; repeated signals can strengthen the system’s guess. These predictions are not proof of a user’s intent and can be inaccurate, but they may still shape what content or ads appear. Defenses include limiting data collection, controlling personalization settings, and treating behavioral predictions as uncertain rather than as confirmed preferences.