Scam.SG
  • Articles
Report a Scam
  1. Home
  2. Scam Prevention
  3. Insecure APIs: How Singapore SMEs Can Protect Customer Data
Scam Prevention

Insecure APIs: How Singapore SMEs Can Protect Customer Data

Editor
17 November 2026
Insecure APIs: How Singapore SMEs Can Protect Customer Data

Summarise this page with:

ChatGPTCopilotClaudeGrokPerplexity
Share this article:

A customer portal can accept the correct password and still reveal the wrong person's records. For a Singapore business handling bookings, orders or membership details, that is a security problem a polished website cannot fix. The weakness may sit in an application programming interface, or API: the connection that lets software request information from another system.

On 12 November 2026, the Personal Data Protection Commission (PDPC) and Cyber Security Agency of Singapore (CSA) issued a joint advisory on data protection risks arising from insecure APIs. It explains how weak credentials, excessive permissions and missing checks can expose personal data, even when requests appear to come from authorised users.

The immediate task for an SME owner is concrete: ask your software provider to demonstrate that each user and integration can retrieve only the customer records and fields they actually need. A successful login is not proof that those boundaries work. This article explains the advisory's prevention lessons; it does not announce a new scam campaign or attribute an unreported breach to any business.

How a Software Connection Can Expose Customer Records

An API is not necessarily a product you bought separately. It may connect your online shop to a delivery provider, a booking page to your customer database, or an application to a membership system. Your staff may use these connections daily without seeing the underlying requests.

The six-page PDPC–CSA advisory separates two checks that businesses often confuse. Authentication establishes who a user or application is. Authorisation decides which records, fields and functions that identity may access. Both matter.

Consider an illustrative online booking service. A customer signs in to see their own reservation. The system should check that the reservation belongs to that customer before returning it. If it checks only that someone is logged in, another person's details could become accessible. This is an example of broken object-level authorisation, a weakness described in the advisory, not a report about a particular Singapore booking company.

The same problem can affect integrations. A delivery service may need a recipient's name, delivery address and contact number. That does not mean it should receive every customer's full account history. The advisory recommends limiting both access permissions and the personal data returned through APIs.

There is another route to exposure: a connection designed for internal use becomes accessible from the internet during a migration or third-party integration. The advisory illustrates a case where a network restriction was removed without replacement safeguards. The business did not realise its internal endpoints had become exposed.

What the Official Advisory Establishes

The advisory is dated 12 November 2026. That date matters: a website's general last-updated label is not evidence of a new incident.

Its examples show why API security deserves a business-level review rather than a quick assurance that the website is online. In one illustration, weak default credentials allowed researchers to query a database containing millions of records. The activity went undetected because it used valid credentials and legitimate API functions. The document presents this as an illustration; it does not identify a newly affected SME.

The advisory also cites earlier PDPC matters involving Marina Bay Sands and Tech in Asia. Those references support the general risk, but they should not be read as announcements of new breaches in November 2026. No fresh loss estimate or victim count is needed to make the prevention case.

For customers, exposed information may make a later impersonation attempt sound more credible. That is a risk to plan for, not proof that every data exposure leads to a scam. Knowing an order detail or membership number does not establish that a caller is authorised to collect money or change an account.

What to Ask Your Software Provider Before the Next Change

Start with an inventory. Ask which APIs access personal data, who owns each connection, which supplier maintains it and whether it is internet-facing. Include old versions and integrations that staff no longer use. The PDPC–CSA advisory recommends regularly reviewing API use and disabling obsolete connections where appropriate.

Then ask for evidence rather than a general security promise:

  • Record boundaries: Has the provider tested that one customer cannot access another customer's records? Ask for a written test result using authorised test accounts and synthetic data.
  • Field boundaries: What information does each connection return? Remove fields that are unnecessary for its stated purpose.
  • Supplier access: Can each external service access only the data and functions it needs? Review permissions when a contract or integration changes.
  • Credential handling: Have default passwords been changed? Where are keys and access tokens stored, and who can revoke them promptly?
  • Change checks: What security tests run before and after a migration, software upgrade or new integration?
  • Monitoring: Who receives an alert when an account retrieves an unusual volume of customer data?

These questions translate the advisory into a procurement and management conversation. An owner need not perform penetration testing personally. The useful outcome is an accountable provider, a documented scope and evidence that fixes were tested.

Do not ask staff to experiment on unrelated websites or retrieve real customers' information to prove a weakness. Testing should be explicitly authorised and confined to the agreed environment. If a customer reports seeing someone else's data, preserve the report and escalate it privately rather than requesting further examples from other accounts.

Practical Controls for Singapore SMEs

Keep permissions narrow

Give staff and connected applications access according to their tasks. An integration that checks a booking's status should not automatically gain permission to export the customer database. The advisory recommends least-privilege access and periodic permission reviews.

Treat administrative accounts separately. Use multi-factor authentication where applicable, particularly for privileged access. Protecting an administrator's login does not remove the need to check authorisation on customer-facing API requests.

Protect credentials and data in transit

The advisory recommends HTTPS with Transport Layer Security for API communications. It also warns against unnecessarily exposing keys and tokens in source code, publicly accessible files or configuration settings.

Ask the provider how it stores credentials securely and how a suspected key can be revoked. Keeping an integration working is not a good reason to leave a compromised credential active indefinitely. Plan the replacement and containment steps with the people responsible for the system.

Look beyond login failures

A stolen credential may produce successful logins. Monitoring only failed attempts misses that possibility. The advisory recommends logs and alerts proportionate to the data and risk, including unusual request volumes, repeated access to different customer records and unexpectedly large retrievals.

Rate limits and export thresholds can reduce the scale of exposure. They are supporting controls, not substitutes for deciding whether a request is authorised in the first place. A slowly leaking system remains insecure.

Recheck after migrations

Include access controls in the release checklist. If a supplier removes a network restriction to connect a new service, ask what will replace it and how that protection will be verified. The advisory specifically recommends risk assessment and alternative safeguards when existing controls change.

For a small business, a short sign-off record is useful: what changed, what data is affected, who approved it, what was tested and who will monitor it. This is a suggested operating practice, not a prescribed legal form.

What to Do If Customer Data May Have Been Exposed

Bring the system owner, software provider and data protection officer into the same incident process. Stop suspicious access, revoke compromised credentials where appropriate and preserve relevant logs. The advisory recommends procedures for containment, investigation and assessment of whether personal data was affected.

Record what is known and what remains uncertain. Which connection was involved? What information could it return? What do the logs show was accessed? Distinguish potential exposure from confirmed extraction. Avoid public assurances that no data was taken before the evidence supports them.

Use the PDPC's breach-reporting guidance to assess notification duties. It states that not every breach must be reported. A breach is notifiable to PDPC if it is likely to result in significant harm to individuals or affects 500 or more individuals. Notification to affected individuals follows the significant-harm requirements, subject to applicable exceptions; scale alone does not mean the same notification rule applies to everyone.

PDPC says notification to the Commission must be made as soon as practicable and no later than three calendar days after determining that the breach is notifiable. That is not a reason to delay assessment: its guidance says containment and assessment should begin immediately after a potential breach is discovered. Get qualified advice if the obligations are unclear.

If customers need to be contacted, give them plain facts, protective steps and a verifiable contact route. Do not include unnecessary personal information in a mass message. A useful notice should help people act without forcing them to trust an unfamiliar payment link or caller.

If You Are a Customer Receiving a Breach Notice

Verify the notice through the business's known website or contact details you already hold. Do not rely solely on a phone number or link inside an unexpected message. Ask what information was affected and which specific action is required.

If you disclosed banking credentials or see an unauthorised transaction, contact your bank immediately through its official fraud channel. If an account password was exposed, change it through the genuine service and change reused passwords elsewhere. These are precautionary steps; an exposed postal address alone does not establish that your banking login was compromised.

For help assessing a suspicious message, number or link, use ScamShield's checking resources or call its 24/7 helpline at 1799. A caller's knowledge of your purchase does not make a request for an OTP or money transfer safe.

Common Mistakes to Avoid

Accepting “the login works” as a security test. The question is also whether that user can access only the correct records.

Assuming the supplier handles everything. Ask who owns configuration reviews, monitoring and incident escalation. A gap between responsibilities can leave an exposed connection unattended.

Treating every request with a valid key as legitimate. The advisory explicitly warns that credentials can be compromised and used through normal API functions.

Publishing conclusions too early. State the evidence and uncertainty. Do not describe a technical weakness as a confirmed scam campaign or promise customers a specific outcome without support.

FAQ

What is an API in everyday business terms?

It is a connection through which software requests or exchanges information. A customer portal, delivery integration or booking system may use APIs even when staff never interact with them directly.

Does a strong password stop every API data leak?

No. Strong credentials help protect identity, but authorisation must also restrict which records and functions that identity can access. The PDPC–CSA advisory discusses failures where authentication works but permissions do not.

Is HTTPS enough to protect customer information?

No. HTTPS protects information in transit. It does not decide whether the recipient should receive a particular customer's record. Access controls, secure configuration, credential protection and monitoring remain necessary.

Can we leave API security entirely to our website vendor?

A vendor can carry out technical work, but responsibilities should be explicit. Ask for the API inventory, permission reviews, testing evidence and incident contacts. The advisory encourages organisations and DPOs to work with IT teams and service providers.

Must every suspected API incident be reported to PDPC?

Not automatically. Assess whether a breach occurred and whether the notification criteria apply. Follow PDPC's guidance on significant harm, significant scale and the relevant notification requirements, rather than assuming all incidents have the same reporting outcome.

What should I do if a caller knows details from my customer account?

Verify the request independently with the business. Accurate personal details do not prove the caller's identity. If the request is suspicious, call ScamShield at 1799; contact your bank immediately if banking information or funds are at risk.

Conclusion

Ask one question at the next software review: can this connection return information to someone who should not have it? Require evidence for the answer, narrow unnecessary permissions and agree who acts when unusual access appears. The PDPC–CSA advisory gives Singapore SMEs a practical starting point without requiring a new breach to justify action.

If a message about your account asks you to disclose sensitive information or transfer money, pause and verify it separately. For scam advice, call 1799, the 24/7 ScamShield Helpline, or visit scamshield.gov.sg.


For Consumer

  • Search a Company
  • Company Directory
  • Whitelist Directory
  • Virtual Office Directory
  • Company Location Map
  • Report a Scam
  • Submit a Review
  • Flag a Business
  • E-Commerce Abuse Reports
  • Articles & Community
  • Scam Statistics
  • View Scam Types

For Business

  • What is TrustScore
  • Claim Your Business
  • Certification Partnership
  • Advertise with Us
  • Submit an Article
  • Work with Us
  • Intelligence Services
  • Singapore Standard Industrial Classification
  • Report E-commerce Abuse

Platform

  • About Scam.SG
  • Watchlist
  • News & Alerts
  • Data Sources
  • Editorial Standards
  • Media
  • Publications
  • Contact Us
  • Sitemap

About Scam.SG

Scam.SG is Singapore's homegrown business trust and anti-scam platform, with authenticity profiles on over 621,000 Singapore-registered businesses. We verify businesses, educate the public on how scams operate, and detect and disrupt scam activity, helping consumers and business associates reduce the risk of falling into a scam. Our analysis uses proprietary algorithms to assess and score Singapore business entities based on publicly available data signals. A lower score does not mean a business is a scam. Visit scam.sg/terminology for definitions of all platform terms.

Business profiles powered by TrustEngine

Disclaimer

Scam.SG is operated by OnScam (SG) Pte. Ltd. We are not affiliated with, endorsed by, or sponsored by any government agency or department. The information provided on Scam.SG (the “Website”) is sourced from publicly available data, including but not limited to ACRA (Accounting and Corporate Regulatory Authority) data from data.gov.sg and other publicly accessible sources. Whilst we strive to ensure the accuracy and reliability of the data presented, we cannot guarantee its completeness or timeliness. Read more at our disclaimer page.


Privacy Policy
Terms & Conditions
Terminology
Disclaimer
Notice & Take Down
Dispute Resolution
Copyright
Sitemap
Scam.SG
© 2026 Scam.SG, operated by OnScam (SG) Pte. Ltd.