Artificial intelligence (AI) plays an increasingly prominent role in how people search for, process and remember information. AI systems, including ChatGPT and CoPilot, are widely used in areas ranging from healthcare to education. These systems serve as accessible sources of knowledge that allow users to obtain information interactively. Although these technologies offer efficiency and convenience, their increasing use also raises questions about their effects on our memory. This thesis investigated whether the perceived source of information (AI versus human) influences memory retention, and whether general trust in AI exacerbates that effect. Fifty-one young adults (18–32 yrs) watched two short educational videos: one narrated by an AI-generated avatar and one by a human speaker. After the presented information and some attention checks, they were immediately tested on their recall of the information via a fill-in-the-blank test. The same recall test happened after a 1-, 2-, or 5-day delay. Participants’ trust in AI was measured with the Propensity to Trust Automation Scale, adapted to AI context. A main effect of source was found; information coming from an AI-avatar was recalled significantly worse than information coming from a human. Forgetting followed the classic curve, but this decay did not differ between AI and human information. Greater general trust in AI showed only a weak, non-significant negative relation to recall, and it did not interact with source. The results imply that by simply knowing the messenger is artificial seems to discourage deep encoding, echoing “Google-effect” concerns about cognitive off-loading.