Source Monitoring
Source monitoring means keeping track of where information came from. Recognizing a claim or the person repeating it is not the same as knowing its original source.
Why it matters online
A forwarded message, search summary, or AI answer can make a claim familiar while obscuring who first supplied it and whether anyone checked it.
How it can be exploited
A misleading claim can gain credibility when a familiar person or tool repeats it. Several repetitions may still trace back to one unverified source.
Recognition signals
- The answer names no original source, but sounds confident.
- An email and its summary are treated as two independent confirmations.
Protective questions
- Who originally supplied this claim?
- Did this tool verify it independently, or just repeat it?
One claim, two messengers
Your friend forwards an article. An assistant summarizes it. You have encountered the claim twice, but you may still have only one source.
That is different from two people independently checking the same event.
What the research supports
Lindsay and Johnson's 1989 experiment used an office picture followed by an accurate or misleading description. Misleading information increased false reports on a yes/no recognition test. A source-classification test did not show that significant increase, although errors remained.
The study concerns human memory. It does not establish that asking an AI for a citation makes the answer correct or prevents phishing.
Trace it, then check it
Find the original source and what it actually says. A source link helps you inspect a claim; it does not certify the claim.
For sensitive requests, verify through a separate known contact rather than the message's own reply details.
The route a claim took to reach you is not proof that someone checked it along the way.
Sources
- The eyewitness suggestibility effect and memory for source (opens in a new tab)
Memory & Cognition · Supports: Lindsay and Johnson (1989): office picture and later description; source classification versus yes/no recognition, with remaining errors.
Related cases
Video breakdownThe Scam Got an AI Concierge
Barracuda described phishing emails carrying instructions aimed at AI inbox assistants. A suspicious request can return through a helpful-looking summary without having been independently checked.
Related concepts
Social Proof
Other people's choices can be useful clues. But several friends forwarding the same offer are not necessarily several people who checked it.
Explore the conceptTrust Transfer
How confidence in a familiar person, brand, or conversation can spill over to a new and insufficiently verified request.
Demo conceptExplore the concept