MS pumping — also called Artificially Inflated Traffic (AIT) — is a fraud in which bots trigger large volumes of one-time-password and verification SMS to number ranges controlled by a colluding party, who collects a share of the termination fee. Operators and enterprises stop it with rate-limiting, phone-number validation, and an ML-driven SMS firewall.
How SMS pumping and AIT actually work
Every public signup form, OTP request, or 'resend code' button is a way to make a company pay to send an SMS. In an AIT attack, a fraudster automates that action thousands of times, pointing the traffic at destination number ranges they control — or that a colluding operator controls. No human ever reads the code. The verification message is the product being sold, and the enterprise that owns the form pays for every one.
OTP and verification messages make up roughly 89% of international A2P traffic, according to messaging-industry analysis published by the Mobile Ecosystem Forum — which is exactly why one-time passwords are the primary attack surface for pumping. The form is universal, the send is billable, and the payout is automatic.
The economics: why pumping exists
AIT is a revenue-share scheme dressed up as traffic. A rogue or complicit operator receives A2P termination fees for messages landing on its ranges; the fraudster who generates the traffic takes a cut. Because the fee is charged per message and the fraud produces no legitimate delivery cost pushback, the incentive is purely volume. The table below shows who benefits at each hop.
| Party | Role in the attack | Incentive |
|---|---|---|
| Fraudster / bot operator | Automates OTP triggers against a target's forms | Revenue-share cut of termination fees |
| Colluding / rogue operator | Terminates the artificial traffic on its ranges | Inflated A2P termination revenue |
| Enterprise (the victim) | Owns the signup or OTP form being abused | None — pays for every fraudulent send |
| Legitimate transit carriers | Carry the traffic in good faith | Ordinary carriage fees, reputational risk |
The macro numbers explain the urgency. Juniper Research projects global losses from SMS-related fraud will reach roughly $71 billion in 2026. AIT specifically is estimated to have peaked near $2.1 billion in 2023 and to be trending down as detection matures — but 'trending down' is a consequence of controls being deployed, not of attackers losing interest.
Detection signals that expose a pumping run
AIT does not look like human behaviour, and that is its weakness. A firewall or fraud model that watches the right signals separates a real user base from a bot farm quickly.
- Conversion collapse — a sudden spike in OTP sends with a near-zero rate of codes actually being entered or validated.
- Destination concentration — traffic clustering on a small set of number ranges, prefixes, or a single foreign operator that has no business reason to see your verification volume.
- Machine-regular timing — inter-request intervals that are too even, or bursts that track a bot schedule rather than diurnal human patterns.
- New-range appearance — verification traffic suddenly landing on ranges your product has never legitimately reached before.
- Cost-per-conversion drift — the metric that ties it together: your spend per successful verification jumps while successful verifications stay flat.
How to stop it: a layered defense
No single control is sufficient. Pumping is stopped by stacking enterprise-side hygiene on top of operator-grade network controls.
- Rate-limit at the form and the API — per IP, per device, per account, and per destination range. Cap OTP resends and add exponential backoff so a single actor cannot generate unbounded sends.
- Validate the number before you send — use a live number-lookup / HLR check to drop invalid, unallocated, or high-risk ranges before they ever become a billable message.
- Add a challenge on suspicious signups — a lightweight bot check on the form removes the automation that AIT depends on.
- Deploy an SMS firewall on the operator side — see BroadNet's SMS firewall, which scores every submission against sender-ID reputation, route reputation, and behavioural baselines.
- Run continuous ML fraud detection — BroadNet's anti-fraud SMS models correlate throughput, content similarity, and recipient-overlap across senders and routes to catch adaptive attacks that static rules miss.
The white-route share of domestic A2P traffic is forecast by Juniper Research to rise from 80.6% in 2024 to 86.5% by 2029 — the direct result of firewalls and registries pushing grey and fraudulent traffic off networks. Pumping shrinks when the wire is monitored.
What enterprises can do this week
You do not need to be an operator to defend yourself. Model your exposure first: use the OTP cost estimator to see what an inflated-traffic spike would cost against your current volumes, then tighten resend caps and add number validation ahead of the send. When you send verification codes, route them through a provider that runs firewall and fraud scoring inline — BroadNet's OTP delivery applies both before a message leaves the platform. If you want the concept in one page, our answer on what AIT is is a good primer to share internally.
Frequently asked questions
What is the difference between SMS pumping and AIT?
They describe the same fraud. 'SMS pumping' emphasises the action — automating OTP or verification sends to pump up billable volume. 'Artificially Inflated Traffic (AIT)' is the industry term for the resulting traffic. In practice the phrases are used interchangeably to mean bot-generated verification messages whose only purpose is to earn termination revenue.
Why are one-time passwords the main target of AIT?
Because OTP and verification messages are roughly 89% of international A2P traffic (Mobile Ecosystem Forum), the OTP form is the most common billable trigger a stranger can reach without authentication. A public 'send code' button lets an attacker generate paid messages on demand, which is exactly the primitive a pumping scheme needs.
Can rate-limiting alone stop SMS pumping?
It helps but it is not sufficient. Rate-limiting slows a single actor, yet distributed attacks spread requests across many IPs, devices, and accounts to stay under per-source caps. Stopping AIT reliably requires rate-limiting plus number validation plus network-side firewall and behavioural fraud scoring that correlate activity across sources.
Is AIT fraud getting better or worse in 2026?
AIT-specific losses are trending down from a roughly $2.1 billion peak in 2023 (Juniper Research) as firewalls and registries spread, but total SMS-fraud losses are still projected near $71 billion in 2026. The category is being contained where controls are deployed and remains costly where they are not.
References
- Juniper Research — Telco fraud & security research: juniperresearch.com
- Mobile Ecosystem Forum — A2P messaging and fraud reports: mobileecosystemforum.com
- GSMA — Messaging security and A2P fraud resources: gsma.com