Security
IoT security hacks: what the 2026 refrigeration attacks teach connected fleets
Supermarket fridges, water plants, and smart TVs share one weakness. Here's how IoT security hacks really happen and the layered defenses that stop them.

Around October 11, 2026, attackers reached into refrigeration controllers at French supermarkets and pushed temperatures to unsafe levels. Food spoiled, and a second wave hit grocers in at least seven more European countries in early November.
The entry points looked ordinary. Like most internet of things hacks, the most damaging IoT security hacks rarely need novel exploits. They need an exposed device and a default password.
How do IoT security hacks actually happen?
IoT security hacks happen when attackers gain unauthorized access to a connected device or its network. They usually get in through internet exposure, default or weak credentials, or unpatched known flaws. Once inside, they use the device to spy, disrupt operations, or pivot to other systems.
Connected devices attract internet of things hacking because they stay on around the clock and often run unmonitored. Many have no screen or interface, so problems can go unnoticed. Updating firmware in the field also takes planning.
IoT has many security considerations, and each one maps to a layer of the stack. Hologram looks at IoT vulnerabilities across the SIM, the cellular module, firmware, physical and wireless interfaces, and the cloud server. Our guide to where IoT vulnerabilities come from breaks down each layer.
In our experience, weaknesses tend to appear in the gaps between layers and the teams that own them. A firmware team may assume the network blocks inbound traffic, while the network team assumes the device rejects bad logins. Attackers look for exactly that kind of handoff.
Attacks also come in two kinds. Targeted attacks go after a specific company, product, or device type. Non-targeted attacks scan the internet for anything reachable and weak, then move on to the next victim.
Known flaws dominate hacking IoT at scale. The Bitdefender 2025 IoT report, produced with Netgear, found that nearly all (99.4%) observed IoT exploits targeted known, fixed Common Vulnerabilities and Exposures (CVEs). That data comes from consumer smart homes, but the takeaway travels well: patch speed matters.
Real-world IoT hacks and what they share
Three cases show how attacks on IoT play out across retail, water, automotive, and healthcare. Each ends with its lesson, and together they show the internet of things hacked through the same few fixable weaknesses.
Supermarket refrigeration attacks (2026)
Around October 11, 2026, attackers remotely manipulated Danfoss refrigeration controls at supermarkets in France. Claroty Team82 has tracked the French supermarket refrigeration attacks since the first wave. Grocery chain Intermarché said 150 of its stores were affected, with temperatures pushed to 50°C (122°F).
French authorities attributed attacks on refrigeration at 700 supermarkets and other food sector victims to Iran's Islamic Revolutionary Guard Corps Electronic and Cyber Center (IRGC-ECC). That attribution is a government assessment, and the exact attack path in France remains unconfirmed.
A second wave on November 2 hit food storage in markets and butcher shops across at least seven more European countries. In the Netherlands, a Danfoss spokesperson told broadcaster NOS that the affected systems were not on secure networks and used a default password.
Censys measured exposed refrigeration controllers in two product families and found the count fell 53% between October 9 and November 5, 2026. That's a drop to 3,124 hosts from 6,695, though exposure alone doesn't prove a device is vulnerable.
The lesson: exposure plus default credentials is a connectivity design problem. Any IoT hack gets easier when a controller is one search away.
Water system PLCs (2023–2024)
Between November 2023 and January 2024, IRGC-affiliated actors known as CyberAv3ngers targeted Unitronics programmable logic controllers (PLCs). The Cybersecurity and Infrastructure Security Agency (CISA) and partner agencies documented the campaign in the CISA Unitronics PLC advisory. The actors compromised at least 75 devices, including at least 34 in the US water and wastewater sector.
These devices either had a default password or no password at all. The advisory does not report any loss of water service.
The lesson: internet of things vulnerabilities in critical infrastructure often come down to basics. A default password on a reachable controller gives attackers an easy way in.
Medical devices
In January 2025, the U.S. Food and Drug Administration (FDA) issued a safety communication about Contec CMS8000 patient monitors, which are also sold relabeled as the Epsimed MN-120. The FDA flagged three vulnerabilities. A remote attacker could take control of a monitor or crash it. The firmware included a backdoor. Once connected to the internet, the monitor began sending patient data outside the care environment. Cybersecurity and Infrastructure Security Agency’s advisory added that all vulnerable monitors on a shared network could be exploited at the same time. The FDA was not aware of any incidents, injuries, or deaths. Its advice was to unplug the device or use it for local monitoring only, with the network turned off.
Medical devices face the same internet of things vulnerabilities as other fleets, with far less room for error. Our overview of medical device hacking risks covers how manufacturers approach them.
The lesson: when people depend on a device, security belongs in the design from day one. That includes the update path, the network design, and the SIM
The most common IoT attack types
Most internet of things attack patterns fall into seven categories, and each maps to a specific weakness. Our breakdown of common ways attackers hijack devices (https://www.hologram.io/blog/4-ways-cyber-attackers-may-be-hacking-your-iot-devices-right-now/) goes deeper, and the table below gives the short version.
| Attack type | How it works | Typical weakness | First defense |
|---|---|---|---|
| Credential attacks and brute force | Attackers try default or common logins at scale | Factory credentials left in place | Unique credentials set at install |
| Exploitation of known flaws | Attackers run public exploits against disclosed bugs | Slow or missing patching | Signed over-the-air updates |
| Botnet recruitment and distributed denial-of-service (DDoS) | Malware enlists devices to flood targets with traffic | Open Telnet ports and default logins | Closed ports and private networking |
| Man-in-the-middle and cellular interception | Attackers intercept traffic | Unencrypted traffic | End-to-end encryption and VPNs |
| Firmware reverse engineering | Attackers extract firmware to find secret | Keys stored in firmware | Secure boot and hardware-based identity |
| Exposed services such as Universal Plug and Play (UPnP) | Devices open ports or services to the internet automatically | Network services on by default | Unused services turned off |
| Physical tampering and SIM theft | Attackers open devices or remove SIMs to reuse them | Removable, unlocked SIMs | SIM locked to its device |
In consumer homes, Bitdefender and Netgear found streaming devices (25.9%), smart TVs (21.3%), and IP cameras (8.6%) were the most frequently targeted. Cellular gives connected fleets a strong foundation, and like any network, it performs best with the right configuration.
How IoT botnets form: from Mirai onward
Mirai dates to 2016, but it remains the clearest case study of how IoT botnets form. It scanned the internet for devices with open Telnet ports. It then tried logins drawn from a preset list of 62 default credentials.
The USENIX Security paper Understanding the Mirai Botnet found that Mirai infected nearly 65,000 devices in its first 20 hours. It then settled into a steady population of 200,000–300,000 infections.
The authors traced Mirai's rise to the absence of security best practices in IoT, and they urged a shift from default-open ports to default-closed. Hologram has noted that Mirai's source code went public about a week after the attack, which let copycats build their own versions.
How AI is changing the attacker's timeline
AI is shortening some steps in an attacker's workflow. In a lab test, Claroty Team82 documented AI reproducing known exploits for a patched Danfoss flaw. The AI agent reached a working exploit in about 28 minutes, for about 6.00 USD in AI usage.
The test had clear limits. A researcher guided the agent from public vulnerability records, so this was not autonomous discovery. There's also no evidence that AI played a role in the real refrigeration attacks.
Claroty suggests the window between disclosure and exploitation may be shrinking. If that holds, internet of things hackers could move from a published flaw to a working attack faster, so patch speed matters.
How to protect connected fleets from IoT hacks
No single layer stops every attack, so strong programs stack defenses across devices, networks, SIMs, and operations. Security is also shared. Customers encrypt their data and secure their devices and servers, and connectivity providers secure their own environments and routing.
Network-level security has a practical edge because you can update it even when device-level security is fixed in the field. Our guide to IoT security best practices covers the full checklist. For teams comparing tools, our roundup of enterprise IoT security solutions covers the options.
Device and firmware layer
Change default credentials at install and give every device a unique password. Add multi-factor authentication (MFA) and role-based access for anyone who manages devices or dashboards.
Secure boot and hardware-based identity make it harder to load tampered firmware or clone a device. Sign every firmware image, build a reliable over-the-air update path, and replace devices that can't take patches.
The National Institute of Standards and Technology (NIST) publishes NIST IR 8259 guidance as voluntary direction for manufacturers. Revision 1, published April 9, 2026, covers foundational activities for IoT product manufacturers.
Network and connectivity layer
Start by removing public internet exposure. If a PLC needs remote access, CISA's advisory recommends putting a network proxy, gateway, firewall, or virtual private network (VPN) in front of it.
The advisory also recommends segmenting those controllers from other networks. The same principle applies to IoT fleets, so keep connected devices segmented from the rest of your network. When you need to reach devices, choose a method for cellular IoT device communication that sends data and commands without exposing them publicly.
Cellular connectivity makes this simpler. A private access point name (APN) keeps device traffic off the public internet, and a VPN encrypts data end to end. Our explainer on private APNs and VPNs shows how they work together.
With Hologram, a private APN with device separation comes standard. Hologram SDN security isolates devices from the internet and from each other with network segmentation and multiple layers of firewalls. Only authorized cellular devices can connect, all inbound connections are authenticated, and 256-bit Advanced Encryption Standard (AES-256) encryption protects traffic.
SIM and fleet visibility layer
Removable SIMs carry their own risk. Thieves have taken SIMs from shared scooters, for example, to use the data plans elsewhere.
International Mobile Equipment Identity (IMEI) locking ties each SIM to its approved device. Hologram pairs it with stolen SIM prevention and behavior detection, and you can see these controls on Hologram's global IoT SIM. Hologram's industrial SIMs also support private APN, VPN, IMEI locking, and AES-256 encryption.
Visibility closes the loop. The Hologram app alerts you when a SIM's location doesn't match its paired IMEI. Watch data usage too, because an unexpected data spike can signal a compromised device.
Regulation is raising the floor
Regulators are setting a higher baseline for connected products. Under the EU Cyber Resilience Act, the European Commission says manufacturers have had to report actively exploited vulnerabilities since September 11, 2026.
Manufacturers must send an early warning within 24 hours and a full notification within 72 hours. The reporting duty covers products already on the EU market. The Act's main obligations apply from December 11, 2027, so teams selling in the EU have time to prepare.
Where to start: audit what’s exposed
The next wave of IoT security hacks will look like the last one unless teams design connectivity to keep devices off the open internet. This week, audit which devices are publicly reachable and which still use default credentials. These devices keep food cold, water flowing, and patients cared for, and the people who depend on them deserve that attention.
FAQs
Which IoT devices are most commonly targeted by hackers?
In consumer homes, Bitdefender found streaming devices, smart TVs, and IP cameras are the most frequently targeted. In industrial settings, internet-exposed controllers and PLCs are common targets.
In what areas are IoT devices vulnerable to cyberattacks?
Many devices ship with default credentials, sit exposed on the internet, and are slow or impossible to patch. Most also lack a screen or interface, so a compromise can go unnoticed.
How do you make IoT devices more secure?
Change default credentials, patch quickly, and remove public internet exposure. Then isolate traffic with private networking such as a private APN, and monitor for anomalies like unexpected data spikes.
