Hostile SIM Cards Can Hijack Phones, EV Chargers, and IoT Gear, Birmingham Researchers Show

A malicious SIM card can take over a smartphone, an EV charger, or an industrial IoT module using commands that are written into the cellular standard itself. Researchers at the University of Birmingham demonstrated the attack at the USENIX WOOT conference on offensive technologies in Baltimore this week, and the list of what a hostile SIM can do is not short. It can read a device's unique identifiers, send messages and place calls from the victim's number, force a connection down from 4G to a less secure 2G network, or shut the device down completely. In one case, the team forced a locked Android phone to open an attacker-controlled website with no user interaction at all.

The attack path runs through a feature called Proactive SIM. It lets a SIM card send a limited number of special commands straight to the device's modem, and one of those commands asks the modem to execute an AT command — the same instruction language used to control and configure modems since the 1980s. AT commands were never designed to be reachable from the SIM side, yet the researchers found the interface is present on a surprising share of real hardware.

Nine of 26 devices exposed the vulnerable interface

The team, led by PhD researcher Tomasz Piotr Lisowski and assistant professor Dr Marius Muench, built a toolkit called CATana to test how devices respond to SIM-originating AT commands. Kristian Covic of the security firm Fuzzware joined them on the project. They tested 26 representative devices: 18 smartphones and eight cellular-connected IoT modules, including hardware commonly embedded in EV chargers, industrial equipment, and connected cars. The device list spanned multiple manufacturers and operating systems.

Nine of the 26 devices executed SIM-originating AT commands, and that interface led to four discovered vulnerabilities. The USENIX paper documents command execution, arbitrary file read, connection downgrades to 2G, and denial of service as the demonstrated impacts. In other words, a compromised SIM stops being a passive identity card and becomes a control channel into the device's modem, and from there into the device itself.

Why IoT gear is the softest target

The researchers argue the risk matters most for exactly the devices this site covers. Industrial equipment, vehicle systems, and routers are usually locked down, with few exposed interfaces and no app stores or browsers for an attacker to poke at. A SIM slot may be one of the only physical ports on the box, which makes the SIM AT interface an unexpected entry vector into an otherwise hardened device.

That framing flips the usual security playbook. Most teams spend their energy protecting network ports, web dashboards, and firmware update paths. Nobody treats the SIM card as an attack surface, partly because the SIM sits in the carrier's supply chain and partly because it is tiny, tamper-evident, and generally trusted. Yet the SIM is a computer in its own right, running an operating system and holding cryptographic keys, and it has a privileged, standards-defined channel straight into the modem.

Dr Muench describes the situation in blunt terms: "The fascinating part here is that the proactive capabilities of a SIM and the resulting attack surface is explicitly defined in the technical specifications for cellular communication, resulting in 'specification-compliant' attacks."

He also notes why the problem has lingered. "Other researchers, cybersecurity experts, and leaked intelligence documents have shown some of the dangers of hostile SIMs before us. Yet, the resulting risks have not been fully mitigated. Potentially, this is because hostile SIMs are not included in most threat models."

Legacy features, modern consequences

Proactive SIM technology is decades old. It was built when the threat model assumed the network, the carrier, and the SIM itself were all trustworthy, and when a phone was simply a phone. The Birmingham group argues many of these features are no longer needed and now create needless risk as the mobile ecosystem has widened to include chargers, meters, vehicles, and factory controllers.

The surveillance angle is worth spelling out. Because a malicious SIM can instruct the modem silently, victim devices can be turned into listening and tracking tools. The researchers demonstrated on recent Android hardware that a hostile SIM could open an attacker-controlled webpage on a locked phone, no tap required — a handy way to phish credentials or drop malware onto the device before the owner even picks it up. The same interface can interfere with connectivity, which raises the possibility of using a hostile SIM to keep a device off the network or push it onto a weaker one.

The team did not stop at publishing. They coordinated with the GSM Association and with affected chip and device makers. Dr Muench says the reports were taken seriously, and key manufacturers shipped software updates and hardened configurations — work that will benefit billions of devices, from phones and connected vehicles to payment terminals, routers, critical infrastructure, and EV charging systems.

The issues are tracked under CVE-2025-48618, CVE-2026-57550, and CVD-2026-0122. Lisowski says the team will keep digging: "The attacks we found only scratch the surface of what is possible with hostile SIM cards."

Who controls your SIM matters

For consumers, the practical takeaway is that a SIM card is security hardware and should be treated as such. A SIM bought from a random outlet, received in the post as a promotional gift, or installed by someone with access to the device is a trust decision, and this research shows the cost of getting it wrong. The same logic applies to eSIMs: the research covered malicious physical SIMs, and the four attacker scenarios it maps out include compromised eSIM provisioning, where the identity is delivered over a data channel rather than a plastic card.

For fleet operators and manufacturers the message is sharper. If your EV chargers, telematics boxes, or industrial controllers hold a SIM, the modem interface on those devices deserves the same scrutiny as the OS and the network stack. Ask the module vendor whether SIM-originating AT commands are enabled, whether the firmware accepts them, and what the device does when the network hands it an unexpected command. A short questionnaire now beats a field recall later.

Some vendors are already moving on this. The disclosure process with GSMA and the chip makers has produced software updates and hardened configurations for key products, which suggests the fix is manageable where vendors take the interface seriously. On the device side, options include filtering or rejecting SIM-originated AT requests in the modem firmware, disabling proactive commands that are not needed for the deployment, and logging any attempt by the SIM to reach modem functions. On the network side, carriers can monitor for SIMs that behave oddly — repeatedly issuing proactive commands, for example — and can flag accounts tied to suspicious hardware. None of this is exotic; the hardest part is deciding that the SIM deserves this attention at all.

There is also a supply-chain angle that operators should not miss. A hostile SIM has to get into the device somehow. It can be swapped in by someone with physical access, shipped as part of a gray-market batch of hardware, or provisioned remotely in the eSIM case through a compromised profile-delivery channel. That puts procurement and logistics teams in the security chain: know where the SIMs came from, who handled them, and whether the activation records match the devices in the field. Inventory hygiene is a boring-sounding control, but it is precisely the control that makes this entire class of attack harder to pull off. The researchers' four attacker scenarios — physical SIM tampering, interception during distribution, malicious insider action, and compromised eSIM provisioning — all depend on a bad SIM reaching a good device.

The full paper, "CATana: On the Dangers of SIM-Originating AT Commands," by Lisowski, Covic, and Muench, is open access and was presented at USENIX WOOT 2026. The university's announcement covers the findings in plain language, and E+T Magazine reported the work earlier this month. The paper page at USENIX carries the abstract and the full text.

None of this means SIM cards are about to become untrustworthy en masse. GSM carriers and vendors responded to the disclosure, patches are shipping, and the attack requires a hostile SIM in the first place — it is not something a random website or a malicious app can pull off by itself. What it does mean is that the industry's oldest identity chip has a fresh list of questions to answer, and that the humble SIM slot, present on billions of devices, deserves a place in threat models it has never occupied before.

For more on the connected-device economy, see the site's ongoing IoT coverage and cybersecurity coverage.