{"paper_id":"2609.27404","title":"When Trust Attracts Fraud: AI and Trust Arbitrage","authors":["Xieyu Yin","Fenghua Wen"],"paper_date":"2026-09-23","paper_type":"theoretical","paper_url":"https://arxiv.org/pdf/2609.27404","page_url":"https://thequant.space/flowcharts/when-trust-attracts-fraud-ai-and-trust-arbitrage/","scores":{"empirical_rigor":2,"math_complexity":7.5,"quadrant":"Lab Rats"},"proposes":{"status":"available","statement":"The paper develops a two‑market signaling model in which generative AI lowers fabrication, verification, and targeting costs, showing how delayed verification creates a trust valley and enables self‑limiting trust arbitrage across markets.","quote":"We develop a two-market signaling model in which generative AI lowers fabrication, verification, and targeting costs. Our contribution is to show how endogenous veriﬁcation can redirect de-ception toward ex ante higher-quality markets.","summary":"Trust can attract fraud when it delays verification. We develop a two-market signaling model in which generative AI lowers fabrication, verification, and targeting costs. When fabrication becomes profitable before verification, claim credibility first falls and later recovers. Across markets…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"Verification perfectly reveals whether evidence is genuine; the receiver accepts verified genuine evidence and rejects detected fabrication.","type":"model","stated":"explicit","quote":"(2) Veriﬁcation perfectly reveals whether the evidence is genuine: the receiver accepts veriﬁed genuine evidence and rejects detected fabrication.","support":"full"},{"assumption":"The receiver does not trade without positive evidence and has zero outside options; fabrication or verification costs are paid when incurred.","type":"model","stated":"explicit","quote":"The re-ceiver does not trade without positive evidence. Outside options are zero, and fabrication or veriﬁcation costs are paid when incurred.","support":"full"},{"assumption":"Assume π_u − (1 − π) d > 0 so the receiver accepts unchecked pooled evidence.","type":"model","stated":"explicit","quote":"Assume πu −(1 −π)d > 0, (3) so the receiver accepts unchecked pooled evidence.","support":"full"},{"assumption":"Functions f (fabrication cost) and k (verification cost) are continuous, strictly decreasing, with f(0) > b and k(0) > B(π), converging to zero as AI capability a → ∞.","type":"model","stated":"explicit","quote":"Assume that f and k are continuous and strictly decreasing, with f(0) > b and k(0) > B(π), and that both converge to zero as a →∞.","support":"full"},{"assumption":"Thresholds a_F and a_V(π) are uniquely defined by f(a_F)=b and k(a_V(π))=B(π).","type":"model","stated":"explicit","quote":"These assumptions yield unique thresholds: f(aF) = b, k(aV (π)) = B(π).","support":"full"},{"assumption":"Targeting either market costs c(a)+ε in addition to fabrication cost f(a), with c(a)>0, c′(a)<0 and c(a)→0.","type":"model","stated":"explicit","quote":"Targeting either market costs c(a)+ε in addition to fabrication cost f(a), where c(a) > 0, c′(a) < 0, and c(a) →0.","support":"full"}]},"datasets":{"status":"none_found","items":[],"known":[],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"none"},"transaction_costs":{"status":"not_applicable","modeled":"n/a","headline_net_of_costs":null,"detail":null,"quote":null,"source":"front_matter","rule":"R3","candidates":{"front_matter":"n/a","evidence_card":"n/a","evidence_card_quoted":false},"meaning":"the paper reports no trading or execution result"},"results":{"status":"available","evaluation_type":"none","findings":[{"finding":"When fabrication becomes profitable before verification, claim credibility follows a non‑monotonic “trust valley,” dropping from 1 to the prior π and later recovering after verification becomes worthwhile.","kind":"theoretical","quote":"Claim credibility drops from 1 to the prior π when fabrication starts and recov-ers only after checking becomes worthwhile. When fabrication becomes proﬁtable before veriﬁcation, cred-ibility follows a trust valley.","support":"full"},{"finding":"Higher prior quality delays verification, creating an interval where only the lower‑quality market checks; if targeting becomes profitable in this interval, deceptive sellers enter the higher‑quality but less vigilant market.","kind":"theoretical","quote":"The veriﬁcation threshold increases with prior quality: ∂aV ∂π = − d k′(aV ) > 0. When targeting becomes proﬁtable inside the cross-market veriﬁcation gap, mobile deceptive sellers enter the higher-quality but less vigilant market.","support":"full"},{"finding":"The inflow of deceptive sellers triggers verification in the higher‑quality market, which then deters further entry, making the trust arbitrage self‑limiting.","kind":"theoretical","quote":"The inﬂow eventually triggers veriﬁcation, which rations further entry. The resulting trust arbitrage is self-limiting: entry can initially reverse the reliability ranking, but it also activates veriﬁcation and deters further entry.","support":"full"},{"finding":"Generative AI can lower fabrication, verification, and targeting costs at different rates, driving the dynamics described above.","kind":"theoretical","quote":"5. Conclusion Generative AI can lower fabrication, veriﬁcation, and targeting costs at diﬀerent rates.","support":"full"},{"finding":"The verification threshold increases with prior quality (∂a_V/∂π = −d k′(a_V) > 0).","kind":"theoretical","quote":"The veriﬁcation threshold increases with prior quality: ∂aV ∂π = − d k′(aV ) > 0.","support":"full"}],"reported_performance":null,"out_of_sample":{"method":"none","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":18,"citations":{"cited_by_in_corpus":0,"by_independent_authors":0,"how_cited":{"used_as_baseline":0,"uses_method_or_data":0,"extends":0,"discusses_findings":0,"mentions":0,"unclassified":0},"independent_papers":[],"scope":"Citations between papers in this corpus only; citations from outside it are not counted."}},"implementation":{"status":"available","public":false,"code_url":null,"note":"No repository presented by the authors as their own was found in the paper. Absence here is not proof that no code exists."},"replication":{"status":"none_found","card":null,"note":"No replication of this paper has been run by thequant.space yet."},"extraction":{"text_source":"fulltext","model":"openai/gpt-oss-120b","version":2,"extracted_at":"2026-11-14T19:39:26Z"},"provenance":{"claims":"Everything here is what the authors state in the paper. Nothing was re-run or reproduced by the platform.","quotes":"Assumptions, datasets, cost detail and findings were extracted by a language model. Every quote is verbatim paper text. Where an item has `support`, \"full\" means every figure in it appears in the quoted sentences and \"partial\" means a figure is stated elsewhere in the paper; items with a figure found nowhere in the paper were discarded.","datasets_known":"datasets.known lists well-known vendors and public datasets by a stable id, found by a dictionary over the full text where the sentence says the data is used; it is independent of the model and covers only names in the dictionary.","flags":"transaction_costs.modeled, results.evaluation_type and results.out_of_sample.method are platform classifications (structured-decision models); see https://thequant.space/score-guide/#penalty-flags","comparisons":"supports / contradicts / qualifies are a language model's reading of two papers' quoted findings, restricted to closely related papers that share no author. They are pointers to read, not verified replications, and coverage is partial.","implementation":"code_url is a repository the paper itself presents as its own (rule-scored, ambiguous cases judged by a model), checked for liveness when it was recorded."},"docs":"https://thequant.space/agents/"}