Fraud & Scams
1 gain2 Traces12 problems
Every measured gain and problem on the record for this topic, newest first.Too few to weigh yet. Updated .
One square per record, darker for stronger evidence:Evidence: 0 high9 moderate6 emerging
15 records so far, too few to weigh yet. Is your study missing? Add it
AI in fraud and scams
A Trace is one record with both a gain and a problem, so it counts on both sides: 3 records show a gain and 14 show a problem. 11 context records are in the record list but not counted, because they report no measured gain or problem.
| Evidence | Gain side | Problem side |
|---|---|---|
| HighStrong study design, backed by at least two independent sources | 0 | 0 |
| ModerateA measured study or an official source | 2 | 8 |
| EmergingEarly, small, unreviewed or reported second-hand | 1 | 6 |
By type of evidence: 6 news reports; 3 controlled comparisons; 2 surveys; 1 regulatory or legal action; 1 systematic review or meta-analysis; 1 commentary or narrative review; 1 set of official statistics. How evidence is scored
15 records so far, too few to weigh yet. A written summary appears once a topic has at least 20 counted records, with at least 5 on each side. Is your study missing? Add it
In a preregistered study, participants classified real versus AI-generated faces and real versus AI-generated voices at rates significantly above chance.
In a survey of 115 Jordanian Islamic bank staff, those using more AI rated money laundering and fraud detection as more effective, partly because they felt more confident using AI outputs.
In a survey of 70 people, respondents generally supported AI-based verification and favoured brands using AI tools to evaluate influencers, though this is opinion, not measured improvement.
UK fraud reports to Cifas' national database reached a record 444,000 last year, up 6% on 2024, with the body saying criminals are using AI to take over accounts.
A review of 223 studies found AI fraud detectors are built for single scam types among 16 studied, generalize poorly, weaken on outdated training data, and report results inconsistently.
In a 529-person experiment, listeners correctly identified speech deepfakes only 73% of the time in both English and Mandarin, and example deepfakes improved results only slightly.
2 records
A descriptive study collected questionnaire responses from 70 people about AI systems that flag manipulated follower counts, artificial engagement and doubtful influencer authenticity. Respondents showed cautious acceptance, with half saying they somewhat trust AI.
RCHUB JOURNAL OF CONTEMPORARY TRENDS IN MANAGEMENT COMMERCE AND ECONOMICS (JCMCE)By June 2026, researchers had tested whether people can tell real from synthetic faces and voices and whether that skill transfers across senses.
ace