125M SMS Exposed: Is SMS Still a Safe Second Factor?

Many people think their phone number is private if they never typed it into a public website. The CSC Telecom incident in the Baltics shows otherwise: your number, and the codes sent to it, pass through companies you have never heard of. Here is what was exposed, why SMS is the weakest common second factor, and a checklist to move important accounts to TOTP or passkeys.
What Was Exposed at CSC Telecom
On 11 November 2026, LSM reported a Cybernews finding about the Latvian unit of CSC Telecom, a provider that relays business SMS to mobile operators. ERR carried the same report. The key facts:
- Cybernews researchers found a publicly reachable ClickHouse database at the end of September 2026.
- It held more than 125 million SMS messages sent from October 2022 to September 2026. More than three million were sent in September alone.
- The exposed data included recipient phone numbers, the names of sending businesses, full message texts, and metadata such as timestamps, message cost and routing information.
- Message texts included authentication codes and temporary passwords, bank transaction alerts, booking details, delivery updates with names and addresses, and medical test results.
- The database updated in real time, so new messages showed up as they were sent.
- Senders named in the reports include Swedbank, SEB, airBaltic, insurers, retailers and the 112 emergency line. Swedbank suspended use of the provider. SEB said it used the service only for marketing messages and was not told about the incident.
A day later, LSM citing LETA reported that CSC Telecom blamed an error in the database access settings. The company said it had closed access, found no evidence of malicious use, and informed CERT.LV. It also said it cannot fully rule out that someone other than the researchers viewed or copied data.
Note the scale. 125 million is a count of messages, not people. One phone number can appear in hundreds of rows, and no count of unique numbers has been published.
How a Number Leaks Without You Publishing It
The reported leak shows the main path. When a bank, airline or shop sends you an SMS, it rarely talks to your mobile operator directly. It uses an SMS gateway or aggregator, which keeps logs: your number, the sender, and often the full text. Each business you deal with may use a different aggregator, and each of those logs is a copy of your number tied to your bank, your airline and your doctor.
So a number that was never posted online can still end up in a phishing list. A log that says "this number gets Swedbank alerts" is better targeting than anything an attacker could scrape. That is why Swedbank told customers not to trust suspicious SMS, calls and emails that ask for passwords or authentication codes, and airBaltic asked for special caution with messages that ask for card data or logins.
Why SMS Is the Weakest Second Factor
The leak adds one more failure mode to a list that was already long.
- The code travels in plain text through third parties. CISA's mobile communications guidance says it directly: "Do not use SMS as a second factor for authentication. SMS messages are not encrypted." An aggregator log with a real-time feed is the same problem without needing access to a carrier network.
- The number can move. NIST SP 800-63B lists an attacker "who has convinced the mobile operator to redirect the victim's mobile phone to them" as a threat. NIST marks PSTN delivery (SMS and voice) as the one "restricted" authenticator and tells verifiers to consider "device swap, SIM change, number porting" before sending a code.
- It is not phishing resistant. NIST says out-of-band and OTP authenticators "SHALL NOT be considered phishing-resistant" because a typed code is not bound to the session. A fake login page can relay it in seconds.
- Recovery often skips the second factor. CISA notes that "some online services may default to SMS during account recovery flows." If a reset code by SMS can replace your password, SMS is your only factor.
TOTP apps fix points 1 and 2: the secret never leaves your device after setup. They do not fix point 3. CISA: "authenticator codes are better than SMS, they are still vulnerable to phishing; only FIDO authentication is phishing resistant." Passkeys and hardware keys fix all four, as long as you also remove the SMS fallback.
Decision Table: What to Use Where
| Account type | Minimum | Target | Remove SMS? |
|---|---|---|---|
| Primary email (it resets everything else) | TOTP | Passkey or hardware key, two enrolled | Yes, including recovery |
| Cloud provider root or org admin | Hardware key | Two hardware keys, stored apart | Yes |
| Domain registrar and DNS | TOTP | Passkey or hardware key | Yes |
| Source control and CI admin | Passkey | Passkey plus hardware key | Yes |
| Password manager | Passkey or TOTP | Hardware key | Yes |
| Bank | Bank app approval | Bank app approval | Only if the bank lets you |
| Mobile carrier account | Account PIN | PIN plus MFA | Not applicable |
| Low-value shopping or forums | SMS is acceptable | TOTP | Optional |
Two rows need comment. Email comes first because nearly every other account resets through it. The carrier row has no second factor of its own: CISA recommends a carrier PIN, which is required for sensitive operations "such as porting a phone number," as a defence against SIM swapping.
Migration Checklist
Work through the table top down, one account at a time.
- Inventory. List every account where SMS is a login factor or a recovery path. Check each service's security and recovery settings, not your memory. A phone number added years ago "for account security" often still works as a reset path.
- Enroll the new factor first. Add a passkey or hardware key, or TOTP where passkeys are not offered. Enroll at least two passkeys or keys (for example a phone and a hardware key) so losing one device is not a lockout.
- Save recovery codes. Store them in your password manager or on paper in a safe place, not in an email draft.
- Remove SMS explicitly. CISA warns that enrolling an app "does not automatically unenroll the account's SMS," which leaves "a weak fallback mechanism." Delete the phone number from the 2FA settings and from recovery options.
- Test the new path. Log out, log in with the new factor, then run the account recovery flow and confirm it does not offer an SMS code.
- Lock the carrier account. Set the port-out PIN or passcode and turn on MFA for the carrier portal.
- Recheck quarterly. Settings drift after device changes and support resets. Put the review on a calendar.
For teams, do the same for shared and break-glass accounts. A cloud root account whose MFA is an SMS to one engineer's phone can be hijacked, and it is lost when that engineer leaves. Hardware keys in a safe and a written recovery procedure fix both. If you self-host shared credentials, Vaultwarden supports several two-factor methods, so the vault does not have to be the weak link.
If You Send SMS Codes Yourself
The other side of the leak is the business that sent the codes. If your product sends OTPs or reset links by SMS, your aggregator sees every one of them. Treat it like any vendor that holds secrets:
- Keep the aggregator in your vendor secret inventory, and know what its logs retain and for how long. Our vendor secret rotation runbook shows a table format that works.
- Make codes short-lived and single use, so a log entry read later is worthless.
- Do not put full reset links or temporary passwords in an SMS. A code that only works in an already open session leaks less.
- Offer passkeys and TOTP, and let users remove their phone number. NIST requires services that use the PSTN to offer at least one alternative authenticator that is not restricted.
Failure Modes to Avoid
- Passkey added, SMS kept. The account is exactly as weak as before. The attacker picks the weakest factor.
- Only one passkey or key. The first lost phone becomes a support ticket, and support often resets by SMS.
- Recovery email on a weaker account. A passkey on your bank does not help if its recovery email still uses SMS 2FA.
- Trusting a text because it looks like your bank. A leak like this one tells attackers which bank or airline texts you, so their message can look routine. Do not follow links in SMS about your accounts. Open the app or type the URL yourself.
Where Akmatori Fits
Account takeover usually starts outside your infrastructure, but response to it is an operations task with steps you can write down. In Akmatori, operators write runbooks that the agent searches before it calls any infrastructure tool, as the architecture docs describe. A runbook for "suspicious login on a privileged account" can list which audit logs to read and which accounts to check, so the first minutes of an incident follow your plan instead of memory.
Conclusion
The CSC Telecom leak needed no SIM swap and no hacked carrier. One misconfigured database held phone numbers, sender names and live one-time codes, because an SMS code passes through too many hands. Move email, cloud, registrar and source control accounts to passkeys or hardware keys, remove SMS from login and recovery, and recheck it every quarter.
If your team wants AI-assisted incident workflows for account takeover response and security hygiene, Akmatori helps SRE teams investigate alerts, coordinate response, and automate safe infrastructure actions. Powered by Gcore for global infrastructure reliability.
