What happened
On October 23, 2026, the Bermuda Police Service (BPS) warned businesses and corporate banking customers about a telephone fraud campaign in which callers posed as fraud investigators from Butterfield Bank. According to the police statement, the calls targeted Bermuda businesses on October 20 and 21 and showed Butterfield's main Bermuda switchboard number, 441-295-1111, on the recipient's caller ID through spoofing (BPS statement via Bernews; The Royal Gazette).
The callers told victims that suspicious international payment activity had been detected on their accounts. Police said the callers used customer account information and fabricated fraud case reference numbers to sound credible, then passed victims to people posing as technical support staff. In some cases victims were told to install remote access software, including AnyDesk, through malicious websites. Police said the criminals then used those methods to make unauthorised international wire transfers, or told victims to move funds into so-called safe accounts (BPS statement via Bernews).
The statement named two websites tied to the campaign, butterfieldcontact[.]web[.]app and butterfieldhelpdesk[.]web[.]app, and said both have been taken down. Police did not say who requested the removals, when the sites were created or how long they were reachable.
Timeline of disclosure
- October 20 to 21, 2026. Calls reach Bermuda businesses, according to BPS.
- October 23, 2026. Bernews publishes an email Butterfield sent to clients. The bank said several clients had reported vishing attempts "in recent days" in which callers pretending to be from Butterfield directed them "to a fraudulent link for live 'support' chat." It added that it will never ask clients to click a link, move money to a safe account or install software (Butterfield email via Bernews).
- October 23, 2026, 2:08 PM Bermuda time. The Royal Gazette reports the police warning, including the two site names (The Royal Gazette).
- October 24, 2026. Bernews publishes the full BPS statement with its list of tactics and precautions (BPS statement via Bernews).
The bank's email and the police statement are independent first-party accounts of the same activity. The bank's email describes the support-chat lure. The police statement adds the dates, the spoofed number, the remote access software and the completed transfers.
What is confirmed and what is not
The police statement confirms the date range, the spoofed number, the pretext of suspicious international payments, the use of remote access software, the two hosted sites and that unauthorised wire transfers took place. It does not say how many businesses were called, how many lost money, the total amount taken or whether any funds were recovered.
It also does not say how the callers obtained customer account information. That gap matters for any response. The details could have come from earlier phishing, a third party or another source, and neither the bank nor the police has attributed them. Butterfield's client email does not describe a compromise of its own systems, and nothing in the public record suggests one.
A pattern Bermuda authorities have already described
The campaign fits a scheme that Bermuda's Financial Intelligence Agency (FIA) documented in a June 26, 2026 alert. Drawing on 42 suspicious activity and transaction reports that Bermuda banks filed between January 2023 and June 2026, the FIA described four stages: initial contact, social engineering, account compromise and fraud execution. Impersonators of banks and technical support services persuaded victims to install remote access tools such as TeamViewer, then moved funds to foreign and domestic accounts shortly after gaining control (FIA alert).
That alert focused on elderly victims. The October campaign applied the same mechanics to businesses and corporate banking customers. Police noted the "potentially serious financial consequences" for businesses that handle substantial payments and international transactions (BPS statement via Bernews). Crime Stoppers Bermuda wrote earlier this month that police and local banks have issued repeated vishing warnings this year, and that a list of numbers linked to fraudulent bank calls that police shared in July was mostly local (Crime Stoppers Bermuda).
Why the displayed number proves nothing
Caller ID spoofing changes the number a recipient sees. BPS put it plainly: "a telephone number displayed on your caller ID is not proof of the caller's identity" (BPS statement via Bernews). The practice is common. TransUnion reported on October 22 that nearly one in five outbound business-to-consumer calls with U.S. mobile carriers was spoofed in the first half of 2026 (TransUnion). That figure covers the United States, not Bermuda, and is context rather than a measure of this campaign.
When the displayed number is the bank's real switchboard, the check many people rely on, comparing the number with the one printed on a card, appears to pass. In this campaign the websites then added a second, visual layer of apparent legitimacy. They were also the route police describe for delivering the remote access software that made the transfers possible.
Why the hosting choice matters
Both named sites sat on web.app, the free subdomain that Firebase Hosting, a Google service, provisions for every project at PROJECT_ID.web.app. Firebase's documentation describes hosting on these project subdomains at no cost, served over SSL by default (Firebase documentation). The bank's name appeared in the part of the address chosen when the site is set up.
For defenders, this choice has two consequences. The pages carried a valid certificate under a parent domain operated by Google. And because no new domain was registered, monitoring that only watches domain registrations for lookalikes of a bank's name would have nothing to flag. Detection has to look at what is published on shared hosting platforms, not just at what is registered.
What this means for banks with corporate clients
Three features of this campaign apply beyond Bermuda. First, the targets were businesses, where the person who answers the phone may be authorised to release international wires. Police specifically asked payment approvers to verify unusual requests independently (BPS statement via Bernews). Second, the attackers split the deception across channels: a spoofed call to create urgency, a fabricated case number to add procedure, and a branded web page to deliver software. Each step looked routine on its own. Third, the web step used a free hosting platform rather than a registered lookalike domain, which places it outside controls built around domain registrations.
For a bank's security and fraud teams, the public signals are few but specific: pages using the bank's name on shared hosting, client reports describing the same script, and support chat or download prompts that the bank does not operate. Those are the points where outside visibility exists.
Where Outtake could have helped
This section is hypothetical analysis, not a description of what happened in this case. It does not suggest that Butterfield uses or lacks Outtake, and it claims no Outtake visibility into this campaign.
Spotting branded support pages on free hosting
The telephone calls happened on a channel outside public monitoring. The two helpdesk pages did not. They were public web pages with the bank's name in their addresses, and Butterfield's email describes them as presenting a "live support" chat. For financial institutions, Outtake describes detecting cloned login portals and "fake support accounts waiting for customers who need help," and says its Agentic Search reads logos, layouts, images and context rather than relying only on exact keyword matches (Outtake Financial Services).
A page on a free hosting subdomain that pairs a bank's name or visual identity with a support chat or a software download prompt is the kind of signal that monitoring could help detect. Whether detection could have come before the calls depends on facts police have not released, including when the pages went live.
Investigating what sits behind the pages
Removing one page can leave the rest of an operation in place. Outtake's financial services page says the hosting, sister domains, social accounts, redirect infrastructure and messaging groups behind a phishing domain "may remain active," and describes a Threat Graph that links those assets in one investigation (Outtake Financial Services). Its brand protection page says agents trace shared hosting, registration patterns, analytics IDs and linked social accounts from a single flagged asset (Outtake Brand Protection).
In this case, useful questions for an investigation would include whether other Butterfield-themed pages existed on the same or other free hosts, which download links the pages served, and whether similar pages imitated other Bermuda banks. Answers could support a fuller picture for the bank's fraud team and could be shared with police.
Supporting removal through the host
Firebase support lists a route for reporting spam, malware or phishing (Firebase Support). Outtake says it sends takedown requests to the registrars, hosts and platforms connected to a campaign in parallel, and lists evidence packaging and platform escalation among its remediation steps (Outtake Financial Services; Outtake Platform). Earlier detection and a well-documented report could support earlier removal. That may shorten the period in which a caller can send a victim to a working page. The hosting provider makes the removal decision.
Giving fraud and contact centre teams the same picture
Butterfield's email shows clients were reporting the calls to the bank. Outtake describes its Intake product as a way to "see the campaign behind" complaints, and says brand threat intelligence can flow into SIEM, SOAR and ticketing tools (Outtake Brand Protection). Matching incoming client reports to known impersonation pages could help a contact centre send warnings with specific details, such as the page names to watch for, and help payment teams recognise a request that follows the pattern. Outtake says it complements, and does not replace, transaction monitoring or fraud detection (Outtake Financial Services).
Trace the Butterfield vishing chain stage by stage
Choose a view, then select a stage. The first view follows the sequence that Bermuda police and Butterfield described. The second groups the same campaign by the party able to act. Reported evidence comes from the cited statements. Possible interventions are hypothetical and do not describe what happened in this case. This section links to no malicious sites and collects no data.
Reported attack chain
1. Account details
Callers arrive already knowing account information
Reported evidence
Police said the fraudsters "reportedly used customer account information and fabricated fraud case reference numbers" to establish credibility (BPS statement via Bernews). The source of that information has not been disclosed.
Possible external intervention
If customer data surfaced in exposed datasets, monitoring those sources could support earlier awareness that a targeted calling list may exist. Outtake lists dark web and exposed data among the sources it covers for financial institutions (Outtake Financial Services).
Control boundary
There is no public evidence that the data came from a leak an outside monitor could see. It may have come from earlier phishing or another source.
2. Spoofed call
The bank's real switchboard number appears on screen
Reported evidence
Calls on October 20 and 21 displayed 441-295-1111, Butterfield's main Bermuda number, and claimed that suspicious international payment activity had been detected (BPS statement via Bernews).
Possible external intervention
No Outtake role is claimed. Caller ID spoofing happens in the telephone network, outside the public web, social, app and ad channels Outtake describes monitoring.
Control boundary
Carriers and telecom regulators control call authentication. The bank's levers at this stage are client education and call-back procedures.
3. Fake case
A fabricated case number and a hand-off to "technical support"
Reported evidence
Callers cited fabricated case reference numbers, then passed victims to people posing as technical support personnel (BPS statement via Bernews). Butterfield told clients the callers directed them "to a fraudulent link for live 'support' chat" (Butterfield email via Bernews).
Possible external intervention
The call script is private, but client reports to the bank are not. Matching those reports against known impersonation pages could help confirm a campaign quickly. Outtake describes its Intake product as a way to see the campaign behind complaints (Outtake Brand Protection).
Control boundary
Matching complaints confirms that a campaign is active. It does not interrupt a call already in progress.
4. Helpdesk site
Branded pages on free hosting deliver remote access software
Reported evidence
Police named butterfieldcontact[.]web[.]app and butterfieldhelpdesk[.]web[.]app, said victims were in some instances told to install remote access software, including AnyDesk, through malicious websites, and said both sites have been taken down (BPS statement via Bernews). web.app is the default project subdomain for Firebase Hosting (Firebase documentation).
Possible external intervention
Public pages carrying a bank's name and a support prompt are a detectable external signal. Monitoring for them, investigating related pages and reporting them to the host with evidence could support earlier removal (Outtake Financial Services; Outtake Platform).
Control boundary
Police did not say when the sites were created or how long they were live. A page created shortly before the calls leaves a narrow detection window.
5. Transfers
Remote access becomes unauthorised international wires
Reported evidence
Police said the criminals used these methods to make unauthorised international wire transfers or told victims to move funds to "safe accounts" (BPS statement via Bernews). The number of victims and the amounts involved have not been disclosed.
Possible external intervention
No Outtake role is claimed at this stage. Removing pages earlier could reduce how many calls reach a working site, but it cannot reverse a transfer.
Control boundary
Payment verification, wire recalls and account recovery belong to the bank, the client's own approval process and law enforcement.
Who can act on each signal
Bank brand and fraud team
The impersonated bank warns clients and collects reports
Reported evidence
Butterfield emailed clients, listed what it will never ask them to do and asked them to report any attempt (Butterfield email via Bernews).
Possible external intervention
Monitoring for brand-named pages and matching them to client reports could give the team specific indicators to share, such as the page names to watch for.
Control boundary
The bank cannot remove pages hosted by a third party. That decision belongs to the host.
Hosting provider
The host decides whether a reported page comes down
Reported evidence
Both named sites ran on Firebase Hosting subdomains, and police said they have been taken down (BPS statement via Bernews; Firebase documentation).
Possible external intervention
An evidence-backed abuse report could support removal. Firebase Support lists a route for reporting spam, malware or phishing (Firebase Support).
Control boundary
The provider decides whether and when to remove content. Police did not say who reported these two sites.
Business payment approvers
Approvers decide whether an unusual transfer goes out
Reported evidence
Police urged employees who start or approve payments to verify unusual requests independently, particularly international transfers and claimed account breaches (BPS statement via Bernews).
Possible external intervention
Warnings that name the actual pretext and page names could make an internal alert more concrete for finance staff.
Control boundary
Approval controls and call-back discipline sit inside the client business, not with any outside monitor.
Police and regulators
Investigators collect reports and describe the pattern
Reported evidence
The Bermuda Police Service issued the advisory and asked anyone affected to report it. The Financial Intelligence Agency's June alert drew on 42 bank reports filed between January 2023 and June 2026 (BPS statement via Bernews; FIA alert).
Possible external intervention
No Outtake role is claimed. A bank's own record of impersonation pages it found and reported remains available to share with investigators.
Control boundary
Attribution, arrests and fund recovery are law enforcement work.
What external monitoring cannot reach
Gaps in the public record
Several facts that would shape any assessment of this campaign are not public. Police have not said how many businesses received calls, how many installed software or how much money left their accounts. They have not said when the two sites were created, how long they stayed online or who reported them. And no one has explained how the callers obtained customer account information, which is the detail that made the calls convincing.
Those gaps limit what can be said about timing. If the pages went live weeks before the calls, external monitoring had a wider window. If they were created minutes before the first call and removed shortly after, the window was narrow, and earlier detection might have made little difference.
Limits of a hypothetical Outtake role
Outtake's published capabilities concern public, external assets: domains, hosted pages, social accounts, ads, apps and exposed data (Outtake Platform). Most of this campaign happened elsewhere.
- The phone channel. Caller ID spoofing occurs in the telephone network. Call authentication is a matter for carriers and regulators, not external brand monitoring.
- Private conversations. The fake case numbers and the hand-off to "technical support" took place on private calls that no outside monitor can see.
- Remote sessions and transfers. Once a victim installed remote access software, the activity moved onto the victim's device and banking session. Endpoint security, payment controls and the bank's fraud systems cover that ground.
- Host decisions. Even a strong, early report depends on the hosting provider choosing to act.
- Reversibility. Removing a page can reduce how many later calls reach a working site. It cannot recover money already sent.
The honest summary is that external monitoring could have addressed one visible step in a five-step chain. That step mattered, because police describe the websites as the route for installing remote access software, but it was not the whole attack.
Frequently asked questions
Was Butterfield's own phone system hacked?
No evidence of that has been published. Police said the callers spoofed the bank's number, which means they manipulated what appeared on the recipient's caller ID (BPS statement via Bernews). Butterfield's client email does not describe a compromise of its systems (Butterfield email via Bernews). How the callers obtained customer account information has not been disclosed.
Why would scammers use web.app pages instead of registering lookalike domains?
Police did not explain the choice. Firebase Hosting gives every project free subdomains on web.app and firebaseapp.com, served over SSL by default (Firebase documentation). An operator can publish a page with a bank's name in the address without registering a domain, so monitoring that only watches new domain registrations would not see it.
What signals were visible from outside the bank?
Two public pages with the bank's name in their addresses (BPS statement via Bernews), and the client reports Butterfield said it had received (Butterfield email via Bernews). The calls, the case numbers and the remote sessions were private and not visible to an outside monitor.
Could Outtake have stopped the calls?
No. Outtake describes detecting, investigating and reporting external threats such as impersonation pages and fake support channels (Outtake Financial Services). Caller ID spoofing, private phone conversations and bank transfers are outside that scope. Earlier action on the support pages could support disrupting one step of the chain.
What should a business do if an employee installed remote access software during a call like this?
Police advise disconnecting the device from the internet where it is safe to do so, contacting the bank on a verified number from a separate trusted device, and not using the affected device for online banking until it has been checked and secured. They also ask victims to report the matter to the Bermuda Police Service (BPS statement via Bernews).
Sources
- Bermuda Police Service statement, published in full by Bernews: BPS Warn Businesses Of Banking Phone Scam, October 24, 2026.
- The Royal Gazette: Public warned over fraudulent bank calls, October 23, 2026.
- Butterfield client email, published by Bernews: Butterfield Warns Of Recent Vishing Attempts, October 23, 2026.
- Financial Intelligence Agency, Bermuda: Phishing and Vishing Scams Targeting the Elderly, June 26, 2026.
- Crime Stoppers Bermuda: Vishing, Spoofing and Fake Bank Calls, October 1, 2026.
- TransUnion: 81% of Suspected Digital Fraud Attacks Target Existing Customer Relationships, October 22, 2026.
- Firebase: Get started with Firebase Hosting and Firebase Support.
- Outtake: Financial Services, Brand Protection and Platform.