EVOPEDIA

← all systems · see it in The System

Business email compromise

A class of cybercrime that uses email fraud to attack organisations (src:wp:email-spoofing): an offender impersonates an employee, supplier or other trusted business party, through a taken-over account, a spoofed address or a look-alike identity, to obtain funds, account numbers or sensitive information (src:cross-gillett-2020); in the FBI's words a scam targeting businesses or individuals who regularly make wire transfers, carried out by compromising email and other channels through social engineering or computer intrusion to make unauthorised transfers (src:ic3-2025). Walked from the defender's and the researcher's side: its kinds, losses and their limits, the people and institutions on each side, the defences and how well they are measured to work, and why the field says it works.

Properties published scienceour proposal

Sourced fields, and our own readings.

born atby 2013 (IC3 statistics from October 2013; Google and Facebook defrauded 2013-2015)
born precisionyear
boundaryInside: the scheme from reconnaissance to the return: the information gathered, the borrowed identity, the tailored request, the recipient's decision as the fraud meets it, the transfer and its laundering through mules, and the groups' division of labour. Outside: the organisation and its employee (who are deceived), the banks, email providers and authentication, law enforcement and IC3, and the training and controls that answer it; they are partners, and their acts are recorded as channels, loops and observations.
nested levelRuns on organisations' payment processes and on people's judgement (Human L5-L7), carried by email; a variant of phishing and of social engineering; a class of cybercrime run by offender groups.
scaletechnological

Measures (33) published science

BEC complaints and reported losses (IC3)2025: 24,768 and $3,046,598,558; 2024: 21,442 and $2,770,151,146; 2023: 21,489 and $2,946,830,270 complaints; US$
BEC losses and complaints reported to IC3, 2014-20182014: $226,000,000 (2,417); 2015: $246,226,016 (7,837); 2016: $360,513,961 (12,005); 2017: $676,151,185 (15,690); 2018: $1,297,803,489 (20,373) US$ (complaints)
Exposed BEC losses, October 2013 to December 2022277,918 incidents; $50,871,249,501 (domestic and international, from complaints, law enforcement and bank filings) incidents; US$
US and international BEC exposed losses (actual plus attempted), June 2016 to July 2019 (FBI)$26,201,775,589 over 166,349 incidents (Cross and Gillett give it as 'over 26 billion' lost) billion US$
Net direct losses as a share of gross BEC losses (estimate)50-75 %
Countries reporting BEC; countries receiving transfers177; over 140 countries
Real estate BEC growth 2020-2022reports +27%; losses +72% % change
Financial Fraud Kill Chain, 2025 (all crime types; no BEC-only figure published)3,900 incidents; $1,163,919,846 attempted theft; $679,013,183 frozen; 58% success (domestic 3,574 incidents, $507,042,623 frozen; international 326, $171,970,560) incidents; US$; %
Kill chain for critical-infrastructure victims, 2025 (all crime types; no BEC-only figure published)655 incidents; $261,451,001 reported; $146,561,094 frozen; 56% incidents; US$; %
BEC complaints from people aged 60 and over (IC3 2025)4,566 complaints; $568,048,472 complaints; US$
BEC losses involving AI reported to IC3 (2025)over 30 million US$
Objectives of sampled BEC attackswire transfer 46.9%; click link 40.1%; establish rapport 12.2%; steal PII 0.8% (3,000 attacks)
Impersonated roles in sampled BEC attacksCEO 42.9%; CFO 2.2%; other C-level 4.5%; finance/HR 2.2%; other 48.1% (recipients: CEO 2.2%, CFO 16.9%, C-level 10.2%, finance/HR 16.9%, other 53.7%)
BEC share of all email (one vendor's customers)less than 1 in 50,000 emails
BEC-Guard precision and false-positive rate98.2%; 1 in 5.3 million emails %; rate
CAPE detector precision over two years in productionabove 80 %
SPF and DMARC adoption among the top million domains (2017)SPF 45.0%; DMARC 4.6% (2015 scans as cited: SPF 40%, DMARC 1%)
Email providers showing a security cue on unverified mail9 of 35 providers
Click rate on spoofed mail with and without a security cue48.9% without, 37.2% with (role play, significant); real-phishing difference not significant
Phishing training effect on clicks (reproduction)no significant effect (p = 0.450; reports p = 0.417); click rate 7.0% easy to 15.0% hard lures p; %
Phishing training trial size (Ho et al.)over 19,500 employees; 10 simulated campaigns over 8 months
Operation Wire Wire (2018)74 arrests; about $2.4 million seized; about $14 million recovered arrests; US$
Operation reWired (2019)281 arrests (167 in Nigeria); nearly $3.7 million seized; about $118 million recovered arrests; US$
UK authorised push payment fraud (2022)£485.2 million in 207,372 incidents (200,643 on personal accounts) GBP; incidents
Free webmail providers used for BECGoogle 68%; Microsoft 17% % of webmail-sent attacks (as cited)
BEC groups profiled; with their own domains; with mail servers20; 16; 13 groups
Price of a commodity remote-access tool used by BEC groups12-69 US$
Targeted against mass campaign yield (as cited)10 times the money from 1,000 times fewer emails ratio
Reconnaissance lead time before a BECweeks or months time
IC3 complaint intakealmost 3,000 complaints per day (all crime types)
Reed warbler nests parasitised by cuckoos, Wicken Fen24% (1985) to 1% (2012); mobbing 90% to 38%; rejecting non-mimetic eggs 61% to 11%
Payroll diversion BEC complaints, January 2018 to June 20191,053 complaints; $8,323,354 reported; average $7,904 complaints; US$
Lain et al. field study sizemore than 14,000 employees over 15 months employees; months

Open contentions (5) open contention

Where sources or readings disagree. Both stay on the map.

Cross-scale patterns seen here (12) our proposal

Shapes that recur when the same kind of problem is solved at different scales, and how each shows in Business email compromise. Each links to every other system that shows it.

Valence: what it trades, and with whom (12) published science

Each exchange with a neighbour, read from both sides: who gains, who pays, and what carries it.

Partners and parts published science

Built from / builds (5)

Sits inside / holds (3)

expressed by (13)

floor filled by (15)

represents (1)

Kinds (11)

Notes published science

Layers

The ways it takes in its world, where they meet, and how it acts back, each from L1 (where it meets the world) up.

business email compromise as fraud researchers, criminologists and security economists describe it: the trusted business relationships it imitates, up to the money and data it diverts (reader's view)

  1. L1: The trusted relationships of business and the mail they travel by: executives and staff, suppliers, customers, lawyers, title companies; email (and phones and meeting apps) used to arrange payments; the organisation's public face
  2. L2: Reconnaissance: names, roles, relationships, processes, travel and writing style gathered from public sources, data breaches and compromised mailboxes, for weeks or months
  3. L3: The borrowed identity: a taken-over real account, a spoofed address, or a look-alike domain or free webmail account under a trusted name
  4. L4: The scheme: a tailored request (pay this transfer, change these bank details, buy gift cards, send W-2 forms) timed and worded for the person who can act
  5. L5: The recipient's decision inside the organisation's payment process: comply, or verify by another channel; runs on the employee (Human) and the organisation's controls
  6. L6: The transfer: funds wired through banks to accounts the fraudsters control, moved on by money mules and second hops, often abroad; recall and freezing contest it here
  7. L7: The return: financial gain (and data for further fraud) for loosely organised groups that divide labour like firms

the boardroom (shared floors)

  1. L4: The scheme: a request built from what reconnaissance found, sent under a borrowed identity to the person who can act; tailored, rare and carrying no malicious payload
  2. L5: The recipient's decision inside the payment process: authority, urgency and routine against verification and sign-off. Runs on the employee (Human's board) and the organisation's decision floor
  3. L6: The transfer and its contest: money wired to accounts the fraudsters control and moved on by mules, against recall, freezing and investigation; since 2022 also to cryptocurrency exchanges, which changes the race the freeze has to run
  4. L7: The return: financial gain and data for other fraud, for offender groups with a division of labour like a firm's

way in: A compromised mailbox

  1. L1: A legitimate account of an executive or supplier, taken over through malware or phishing
  2. L2: Access gained, often weeks or months before the fraud
  3. L3: Conversations, contacts and real invoices read; genuine documents altered with new bank details

way in: Authority: the boss's name

  1. L1: The organisation's hierarchy of executives and staff
  2. L2: A request in the name of the CEO or another senior person; society trains people not to question authority
  3. L3: The wish to do the job well and please management
  4. L4: Compliance with what looks like a direct order

way in: Controls: verification, sign-off and training

  1. L1: Governance processes to validate all payments consistently
  2. L2: Verification by a second channel (a phone call to a known number) or two-factor authentication for changes to account details
  3. L3: Several people sign off large transactions; email money requests always queried by other means
  4. L4: Awareness training (measured effects small or null in phishing trials) and goals that can pull against compliance; reporting: employees with a report button as a collective detector (shown for link phishing)

way in: Detection met on the way: authentication and filters

  1. L1: SMTP carries no built-in sender authentication
  2. L2: SPF, DKIM and DMARC check the sending domain; adoption low (DMARC 4.6% of top domains in 2017); headers only
  3. L3: Filters and detectors: bulk filters miss tailored mail; detectors that learn each organisation's history reach high precision

way in: Law enforcement

  1. L1: Victims' complaints: IC3 averages almost 3,000 a day
  2. L2: Analysts assign crime type and adjusted loss; figures are a point-in-time assessment
  3. L3: Investigation follows the money: bank secrecy lifted, email providers asked for records
  4. L4: Coordinated operations: Wire Wire (2018), reWired (2019)
  5. L5: International cooperation: European Investigation Orders, Joint Investigation Teams, mutual legal assistance

way in: Money mules

  1. L1: People reached by romance and work-at-home scams
  2. L2: Recruited, at times unwittingly, as 'financial agents'
  3. L3: Their accounts receive the victim's money for a small commission; across online fraud generally, online and international accounts and crypto wallets are favoured, and mules are recruited for identity cards to open them (Bekkers)
  4. L4: Funds moved on, often abroad (Hong Kong and China the main 2022 destinations; the UK an intermediary stop, then Mexico and Singapore), or straight to a cryptocurrency exchange or its custodial bank
  5. L5: The paper trail broken; lost funds hard to trace

way in: Public organisational information

  1. L1: Organisational information freely available online; social networks show relationships between employees
  2. L2: Gathered as open-source intelligence: employee names, email addresses, roles (the three top-scoring reconnaissance techniques across twenty groups); lead-generation software
  3. L3: A profile of the target: culture, relationships, how staff talk to each other, travel movements

way in: Recall and freezing

  1. L1: The victim discovers the transfer; time is of the essence
  2. L2: The victim's bank asked to recall the funds; a complaint filed with IC3
  3. L3: The IC3 Recovery Asset Team starts the Financial Fraud Kill Chain with banks and field offices
  4. L4: The recipient bank freezes the account if the money is still there
  5. L5: Second hops and international transfers followed (FinCEN Rapid Response Team, legal attachés, foreign police)

way in: Routine: the expected invoice

  1. L1: Suppliers in other time zones; invoices commonly settled by email
  2. L2: An invoice within the bounds of usual expenditure, with the hallmarks of legitimate business; or a spoofed 'employee' asking HR to move a salary deposit (payroll diversion), a junior and mundane request
  3. L3: A genuine invoice with altered bank details, sent in the real thread; in real estate, a change of payment type (cheque to wire) or of receiving account
  4. L4: Hard to tell from real mail, since payment is a normal topic with vendors

way in: Sender identity: addresses, domains and accounts

  1. L1: Domain names and free webmail accounts (Google 68%, Microsoft 17% of webmail BEC, as cited)
  2. L2: A look-alike domain registered (sixteen of twenty groups registered their own), or a trusted address spoofed
  3. L3: A From line, display name and reply-to that read as the trusted party

way in: Sight: the message on a screen

  1. L1: Light from a screen, often a phone
  2. L2: The eye reads what the mail client shows
  3. L3: Many readers do not display the sender's address; display name and reply-to differ unseen
  4. L4: Look-alike letters pass at a glance; few providers add a warning cue, which helps in role play and less in real tests; warnings on emails effective in a 15-month field study

way in: The bank's checks on the payment

  1. L1: The payment system: wire transfers, easy to make and hard to reverse
  2. L2: Confirmation of Payee (UK): the name on the receiving account checked when a new payee is set up
  3. L3: Fraud checks in the payment journey; customers can be coached by fraudsters past them
  4. L4: Who bears the loss: reimbursement codes cover consumers and micro-enterprises, not larger businesses
  5. L5: Banks detect and stop or revert some BEC transfers

way in: Tools: commodity malware, lead software, language models

  1. L1: Hacking forums and online markets; legitimate marketing lead-generation software
  2. L2: Off-the-shelf remote-access tools and keyloggers bought cheaply (one for $12 to $69)
  3. L3: Fluent text in many languages from AI language models, after people learned to spot machine translation

way in: Urgency: time pressure

  1. L1: The business clock: requests near the end of the business day
  2. L2: Words such as 'urgent', 'request', 'important'
  3. L3: Under time pressure, a decision strategy with less reasoning
  4. L4: Less thinking through, less questioning

way out: Data released

  1. L1: The data used for identity theft and false tax returns
  2. L5: HR or finance staff send W-2 forms or employee lists

way out: Funds moved on

  1. L1: Funds abroad or converted at a cryptocurrency exchange, the trail broken
  2. L3: Mules move the funds on for a commission
  3. L6: Money flow managed by a dedicated role

way out: The payment made

  1. L1: Funds arrive in an account the fraudsters control
  2. L3: The organisation's bank sends the wire
  3. L5: The employee approves the payment or the change of bank details

way out: The request sent

  1. L1: The message in the inbox of the person who can pay or release data
  2. L3: Sent under the borrowed identity, from the taken-over account or a look-alike address
  3. L4: The tailored request composed

way out: The victim's report

  1. L1: The report reaches the bank and IC3
  2. L2: A recall requested from the bank and a complaint filed
  3. L5: The fraud noticed (irregularities, a missing deposit, a supplier asking for payment)

Sources (21) published science

Email spoofing (business email compromise section) (Wikipedia, via Wikimedia Enterprise snapshot 2026-11-01)
Cross C, Gillett R (2020) Exploiting trust for financial gain: an overview of business email compromise (BEC) fraud. Journal of Financial Crime. doi:10.1108/jfc-02-2020-0026 (open copy via OpenAlex)
Fernández Lareo N (2024) Characterization of Business Email Compromise Threat Actors. Thesis, Universitat Politècnica de Catalunya (UPCommons; sections read: introduction, impact, historical evolution, types of attacks, discussion: final matrix, possible connections, evolution, geography and organization, malware usage). Leans on industry threat-intelligence reports (Agari and others) (behind its links)
Federal Bureau of Investigation, Internet Crime Complaint Center (April 2026) 2025 Internet Crime Report (sections read: director's introduction, complaint highlights, IC3 Recovery Asset Team and Financial Fraud Kill Chain, three-year complaint and loss tables, elder fraud tables, AI section, Appendix B definitions, Appendix C notes on the data) (behind its links)
Cidon A, Gavish L, Bleier I, Korshun N, Schweighauser M, Tsitkin A (2019) High Precision Detection of Business Email Compromise. 28th USENIX Security Symposium (open proceedings PDF; sections read: abstract, introduction, 2 statistics and background, 4 design choices on rarity, evaluation summary, 6 evasion) (behind its links)
Papathanasiou A, Liontos G, Liagkou V, Glavas E (2023) Business Email Compromise (BEC) Attacks: Threats, Vulnerabilities and Countermeasures, A Perspective on the Greek Landscape. Journal of Cybersecurity and Privacy 3(3). doi:10.3390/jcp3030029 (sections read: introduction, literary review, role of social engineering, non-technical and technical methods, tracing the money route, procedures after a BEC attack, results, future work, conclusions) (behind its links)
Wassermann S, Meyer M, Goutal S, Riquet D (2023) Targeted Attacks: Redefining Spear Phishing and Business Email Compromise. arXiv:2309.14166 (preprint; sections read: abstract, introduction, terminology and Table 1, targeted attacks and goals, social-engineering principles, conclusion) (behind its links)
Anderson R, Barton C, Böhme R, Clayton R, Gañán C, Grasso T, Levi M, Moore T, Vasek M (2019) Measuring the Changing Cost of Cybercrime. WEIS 2019. doi:10.17863/cam.41598 (sections read: abstract, bank fraud and authorised push payment, 3.10 Business Email Compromise, summary table) (behind its links)
FBI Internet Crime Complaint Center (9 June 2023) Business Email Compromise: The $50 Billion Scam. Public Service Announcement I-060923-PSA (behind its links)
FBI Internet Crime Complaint Center (10 September 2019) Business Email Compromise: The $26 Billion Scam. Public Service Announcement I-091019-PSA (read in full, footnotes included) (behind its links)
Brabec J, Šrajer F, Starosta R, Sixta T, et al. (2023) A Modular and Adaptive System for Business Email Compromise Detection. arXiv:2308.10776 (sections read: abstract, parts of sections 4 and the efficacy evaluation) (behind its links)
Hu H, Wang G (2018) Revisiting Email Spoofing Attacks (end-to-end measurements on 35 email providers and user studies, N = 913). arXiv:1801.00853 (sections read: abstract, introduction, adoption of SMTP extensions, user study 2 results) (behind its links)
Rozema AT, Davis JC (2026) Anti-Phishing Training (Still) Does Not Work: A Reproduction of Phishing Training Inefficacy Grounded in the NIST Phish Scale. WWW '26. doi:10.1145/3774904.3792467 (sections read: abstract, introduction, results H2) (behind its links)
Ho G, Mirian A, Luo E, Tong K, Lee E, Liu L, Longhurst C, Dameff CJ, Savage S, Voelker GM (2025) Understanding the Efficacy of Phishing Training in Practice. IEEE Symposium on Security and Privacy. doi:10.1109/sp61157.2025.00076 (abstract only; no open full text reached) (behind its links)
Braithwaite J (2024) 'Authorized Push Payment' Bank Fraud: What Does an Effective Regulatory Response Look Like? Journal of Financial Regulation. doi:10.1093/jfr/fjae006 (sections read: abstract, introduction, Confirmation of Payee and the Contingent Reimbursement Model Code, footnotes on business accounts and moral hazard) (behind its links)
Thorogood R, Davies NB (2013) Reed warbler hosts fine-tune their defenses to track three decades of cuckoo decline. Evolution 67(12):3545-3555. doi:10.1111/evo.12213 (PMC4209118) (behind its links)
Lain D, Kostiainen K, Čapkun S (2022) Phishing in Organizations: Findings from a Large-Scale and Long-Term Study. IEEE S&P 2022; arXiv:2112.07498 (behind its links)
Stajano F, Wilson P (2009) Understanding scam victims: seven principles for systems security. University of Cambridge Computer Laboratory Technical Report UCAM-CL-TR-754 (refereed version: Commun ACM 54(3):70-75, 2011, doi:10.1145/1897852.1897872) (behind its links)
Lazarus S, Whittaker JM, McGuire MR, Platt L (2023) What do we know about online romance fraud studies? A systematic review of the empirical literature (2000 to 2021). Journal of Economic Criminology 2:100013; doi:10.1016/j.jeconc.2023.100013 (sections read: introduction, method, results, discussion of offender and victim studies, gaps and SPTM critique) (behind its links)
Dixit T, Liu M, Riederer JM et al. (2026) Categorical identity signatures can reduce host error rates during brood parasitism. PLoS Biol. doi:10.1371/journal.pbio.3003667 (open access; abstract and discussion passages read) (behind its links)
Lazarus S, Button M, Adogame A (2022) Advantageous comparison: using Twitter responses to understand similarities between cybercriminals ('Yahoo Boys') and politicians ('Yahoo men'). Heliyon 8:e11142. doi:10.1016/j.heliyon.2022.e11142 (sections read: abstract, introduction passage on Yahoo Yahoo and BEC) (behind its links)