SilentAir Hijacks Asian Telecom Gear to Steal SMS 2FA Codes

Introduction
A previously unidentified intrusion set has been operating from inside the mobile networks of Asian telecommunications operators, using compromised equipment to read the calls and text messages of a small, deliberately chosen group of people, most of them located in Taiwan. The group was named SilentAir this week by Enea, a telecom and cybersecurity software company headquartered in Stockholm. Enea assessed with high confidence that SilentAir is a state-level actor, a judgement that shifts the story from ordinary cybercrime into the realm of intelligence collection.
For security and engineering teams, the significance is not merely that another espionage cluster has been catalogued. It is where the collection happened: inside carrier infrastructure, at the layer where SMS one-time passcodes and voice calls are processed and routed. That is a position from which an attacker can read authentication codes without ever touching a victim's handset.
What Enea actually disclosed
The public details are narrow but consequential. According to Enea, SilentAir relied on hacked equipment sitting inside Asian mobile operators rather than on malware implanted on subscribers' devices. The company traced the group's attack sources to at least six countries across Asia. The collection was selective, aimed at specific individuals rather than broad population-level harvesting, and the affected users were predominantly in Taiwan.
Several important questions remain unanswered in the public record. Enea has not publicly identified the individual carriers whose equipment was compromised, nor the six countries in which attack sources were observed. The duration of the campaign, the number of individuals whose communications were intercepted, whether content was stored or forwarded in real time, and which government or governments may be responsible are all details that are not yet public. Readers should treat any specific attribution beyond Enea's high-confidence state-level assessment as speculation until more is disclosed.
Why interception at the carrier layer is so damaging
Most defensive guidance treats SMS one-time passcodes as a weak but tolerable second factor. SilentAir illustrates the sharper version of that problem. When an adversary controls or can manipulate equipment inside the operator's network, the usual assumptions collapse at once.
- No endpoint compromise is required. The target's phone can be fully patched, unrooted and free of malicious apps, and the interception still works.
- It is device- and platform-agnostic. Feature phones, iOS and Android handsets are all equally exposed, because the weakness sits upstream of the handset.
- It is largely invisible to the victim. Messages and calls arrive normally; there is no obvious artefact for a user or a helpdesk to notice.
- It captures more than codes. Intercepted voice traffic can expose spoken one-time codes, voicemail access, and the content of sensitive conversations, not just the numeric factor.
The targeting pattern matters as much as the technique. Selective interception of named individuals is characteristic of espionage rather than financial fraud. That suggests the objective was intelligence about specific people, and it means the population at risk is likely small, high-value and chosen for their access rather than their account balances.
Who should be paying attention
The most exposed population is anyone in Taiwan whose communications carry political, diplomatic, defence, telecom or civil-society sensitivity, along with the organisations they work for. But the blast radius extends further. Enterprises with staff, executives, journalists or contractors travelling through the affected region should assume that SMS-delivered codes sent to those numbers may have been observable. Organisations that still rely on SMS as the primary second factor for email, VPN, cloud consoles or administrative access carry the same structural exposure wherever their users roam.
Telecom operators across Asia face a separate and more direct problem: their core network equipment and the third-party access paths into it are now confirmed as an active attack surface, not a theoretical one. The absence of named operators in the public reporting does not mean other carriers are unaffected; it means the investigation has not been made public.
Practical mitigation: what to do now
Defenders cannot fix a carrier's network, but they can reduce what interception is worth.
- Retire SMS as a high-assurance factor. Move privileged and executive accounts to phishing-resistant authentication such as hardware security keys or passkeys. App-based push is an improvement over SMS but remains phishable, so treat it as a middle tier rather than the destination.
- Re-map your authentication tiers. Identify every system where SMS is the only second factor for administrative, financial or email access, and prioritise those for migration first.
- Hunt for the downstream effects. Intercepted codes are used to reset passwords and enrol new authenticators. Alert on MFA enrolment changes, unusual password resets, impossible-travel sign-ins and carrier-side account changes for high-value users.
- Change travel guidance. Advise staff handling sensitive material in the region to avoid discussing it over cellular voice or SMS, and to use end-to-end encrypted messaging instead.
- For operators: harden the core. Inventory core network elements and management interfaces, restrict and log vendor remote access, segment network equipment from general IT, apply available patches, and monitor signalling and provisioning anomalies that do not match normal operational patterns.
Grivyonx Expert Analysis
SilentAir is a useful corrective to a habit the industry has fallen into: treating multi-factor authentication as a binary control that is either on or off. What Enea's finding shows is that assurance depends on the path the factor travels. A code generated on a handset and verified by a server is strong; the same code pushed through a compromised carrier core is only as trustworthy as the least-audited box in that network. The practical consequence is that organisations need to reason about authentication assurance per channel and per user population, not per policy checkbox. It also reinforces that telecom infrastructure security is enterprise security: the third-party access paths and management planes inside carriers are, functionally, part of every customer's identity perimeter. Expect copycat activity and renewed scrutiny of core-network visibility across the region, and expect regulators and enterprise customers to start asking operators questions they have not previously had to answer in public.
Conclusion
SilentAir is still thinly documented, and much of what matters, including the identity of affected operators and the full scope of victims, is not yet public. What is already clear is the shape of the threat: a suspected state-level actor achieved persistent access to the routing layer of Asian mobile networks and used it to read the communications of selected individuals, chiefly in Taiwan. The defensive response is not a single patch but a rebalancing of trust away from channels that can be intercepted upstream, combined with serious hardening of carrier infrastructure itself. Teams that treat SMS as a convenience factor rather than a security boundary, and that instrument their identity systems for the aftermath of code theft, will be far better positioned than those that wait for the next disclosure. Grivyonx supports organisations navigating this shift through AI-driven automation, DevSecOps practices and cloud security engineering that reduce reliance on brittle, interceptable authentication paths.
- #SilentAir
- #telecom security
- #SMS 2FA
- #Taiwan
- #state-sponsored
- #Enea
- #mobile networks
- #espionage



