The 2025 AI Scandal



Your Smart Home Is Watching You Back
13.6 billion attacks on consumer IoT devices in 2025. 29 attack attempts per day on the average connected home. Here’s what your devices are actually doing — and a practical playbook to stop it.
The average connected home now has 22 IoT devices and faces nearly 30 cyberattacks every single day — triple the rate from 2024. One in three data breaches now involves an IoT device. And 20% of smart home devices are still running factory-default passwords right now, in 2026.
This isn’t about paranoia. It’s about knowing what’s actually happening on your network, understanding which risks are real versus hyped, and having a concrete checklist that takes a weekend to implement. That’s what this article is.
I’ll be honest about where this article starts: with a problem in the original draft I was asked to improve. It referenced a dramatic event called the “Gemini Hijack” as the central case study — a March 2025 incident supposedly affecting two million U.S. households. I couldn’t verify it. After checking Wired, Reuters, Google’s security bulletins, and the AI Incident Database, there’s no such event on record.
That matters, because if you’re trying to protect your home network, you need accurate threat intelligence — not invented horror stories that muddy the water. So I rebuilt this from verified data. And here’s the thing: the real numbers are alarming enough on their own.
What’s Actually Happening to Smart Homes Right Now
The smart home market hit $174 billion globally in 2025, per Statista — with household penetration at 77.6% in the U.S. Amazon leads with 34% device penetration through Alexa, Google Nest commands 27%. That’s a lot of microphones and cameras living in people’s bedrooms, kitchens, and nurseries.
The security picture underneath that growth is genuinely troubling. A joint report from NETGEAR and Bitdefender analyzed 6.1 million households and 13.6 billion IoT attacks. Their conclusion: residential networks are no longer peripheral targets — they’re central to global cybercrime operations. The biggest threats they identified weren’t exotic AI exploits. They were boring, preventable failures: default passwords, unpatched firmware, no network segmentation.
That’s the part that gets me. Years of security researchers screaming about default credentials. And 20% of devices are still running factory defaults in 2026. Still. The McDonald’s hiring bot breach that made national news in 2025? The entry point was literally the password “123456” on a six-year-old test account. Not AI. Not a sophisticated zero-day. A password a toddler could guess.
Five Ways Your Devices Are Leaking Data Right Now
Before the defensive playbook, you need to understand the attack surface. Not theoretically — specifically.
Default Credentials
Factory passwords like “admin/admin” or “1234” are published in device manuals — which are publicly searchable. Attackers don’t guess. They look them up. 80% of IoT devices are reportedly vulnerable to a wide range of attacks, with default credentials as the leading entry point.
Unpatched Firmware
Your smart thermostat from 2022 may never have received a security update — and the manufacturer may no longer be shipping them. Critical vulnerabilities can sit unpatched for years. IoT devices in Europe faced an average of nearly 70 attacks per week in 2025.
Excessive App Permissions
That fitness band doesn’t need access to your contacts. That doorbell app doesn’t need your calendar. But most users accept all permissions without reading them — and the data flows to vendor servers, ad networks, and in some cases third-party data brokers.
Flat Network Architecture
When your smart bulb and your laptop are on the same network segment, a compromised bulb is a foothold to everything. Most home routers ship with no isolation — everything on one flat network by default.
Cloud Storage Defaults
Ring camera footage, Nest recordings, smart speaker audio snippets — by default, most of this goes to vendor cloud servers. Data breaches affecting major providers have exposed billions of records containing Wi-Fi credentials, device IDs, and household geolocation data.
AI-Assisted Botnets
2025 saw a documented resurgence of Mirai-style botnets, now enhanced with AI-driven scanning to identify vulnerable devices faster. These don’t just steal your data — they weaponize your router and cameras to attack third-party infrastructure. A 22.2 terabit-per-second DDoS attack in 2025 was fueled partly by compromised home IoT devices.
What the Data Actually Says About Privacy
The survey numbers here are striking. 31% of smart home users cite privacy as a top concern, per 2025 adoption data. A separate study found 60–70% of consumers express discomfort regarding how their connected devices collect data — yet adoption continues to climb. People feel uncomfortable and keep buying. That’s not irrational; it’s the same calculation we make about smartphones. Convenience wins until the personal cost becomes undeniable.
What does that data collection actually look like in practice? Smart speakers buffer audio continuously around their wake words — meaning they’re processing ambient sound even when you haven’t invoked them. Smart TVs track viewing patterns and, in many cases, use automatic content recognition (ACR) to identify and log everything displayed on screen, including content from external HDMI sources. Fitness trackers correlate sleep patterns, heart rate variability, and activity data that can infer stress levels, health conditions, and daily routines.
The most documented real-world privacy breach pattern in 2025 wasn’t a dramatic hack. It was the quiet accumulation of behavioral data — occupancy patterns, voice recordings, consumption habits — flowing to vendor servers and, through them, to advertising partners. This is legal, disclosed in terms of service that nobody reads, and deeply invasive.
Three Smart Home Security Failures Worth Knowing About
Rather than invented case studies, here are three documented incidents that illustrate the actual threat landscape.
The Mirai Resurrection (Early 2025)
Cybersecurity experts documented a significant resurgence of Mirai-style botnet activity in early 2025. The new wave targeted home routers and IoT devices to build massive botnets — most homeowners had no idea their hardware was participating. SecureIoT’s 2025 analysis described this as “Mirai Resurrection,” with AI-enhanced scanning enabling faster device enumeration than the original 2016 attacks.
The record-breaking 22.2 Tbps DDoS attack documented in the NETGEAR/Bitdefender report was, in part, fueled by this kind of residential IoT compromise. Your thermostat isn’t being hacked to steal your schedule. It’s being drafted into an army attacking someone else.
Smart City Infrastructure Attacks (Mid-2025)
Mid-2025 saw coordinated attacks against smart city infrastructure across multiple European and North American cities. Traffic management systems, public surveillance cameras, and utility management devices were compromised — attackers remotely disabled systems controlling traffic lights, public transport networks, and emergency response units. This matters to home users because the same categories of devices (cameras, smart sensors, network-connected controllers) exist in residential settings, running similar firmware with similar vulnerabilities.
Baby Monitor and Home Camera Compromises
The most viscerally disturbing documented smart home incidents involve cameras in private spaces. Multiple families have reported hearing strangers’ voices through compromised baby monitors. Lawsuits have documented people receiving death threats and abuse through hacked in-home cameras. SecureIoT’s 2026 comprehensive guide documents smart lock compromises leading to physical break-ins. These aren’t theoretical. They’re in court filings.
Ten Things to Do This Weekend
I’ve ranked these by impact-to-effort ratio. Do them in order if you can. If you only do three, do steps 1, 2, and 3 — they cover about 70% of your actual risk.
- Change every default password. Today.Use a password manager like Bitwarden (free, open-source) or 1Password. Generate a unique strong password for every device and every account. Enable account lockout after three failed attempts where the device supports it. This single step eliminates the #1 entry vector for IoT breaches. Not exciting. Extremely effective.
- Segment your network — IoT devices separate from everything else.Create a dedicated VLAN or guest network for all IoT devices. Your smart bulbs, thermostats, cameras, and speakers go on a separate subnet that can’t communicate with your laptops, phones, or NAS drives. In your router admin panel, assign all IoT devices to the isolated network. Most modern routers (including consumer ones) support this. A compromised bulb can no longer reach your banking sessions.
- Audit every app permission on your phone. Revoke what’s excessive.Go through your smart home apps and ask: why does this app need location access? Why does it need my contacts? The fitness band tracker doesn’t need your microphone. The smart plug app doesn’t need your photos. Strip permissions to the minimum required for the device to function. On iPhone, use Settings → Privacy & Security. On Android, Settings → Privacy → Permission Manager.
- Run a full firmware update sweep.Check every device manufacturer’s app and website for pending updates. Set up automatic updates wherever possible — tools like Ubiquiti’s UniFi can automate this for whole ecosystems. According to SentinelOne’s 2025 threat report, timely firmware updates reduce breach probability by approximately 65%. That’s not a minor improvement.
- Install a VPN on your router (not just on your laptop).Router-level VPN coverage protects all devices simultaneously, including those that don’t have native VPN clients. Flash DD-WRT firmware on compatible hardware, then configure NordVPN or Mullvad at the router level. Test afterward at ipleak.net. If you see any leaks, redo the configuration — don’t skip this verification step.
- Set up Pi-hole for network-wide ad and tracker blocking.Pi-hole runs on a Raspberry Pi and acts as a DNS sinkhole for your entire network, blocking known ad servers and tracking domains before they can phone home from your devices. It’s free, open-source, and takes about an hour to configure. After installation, you’ll be able to see exactly which of your devices is making outbound connections — and to where. Genuinely eye-opening.
- Move cameras to local storage where possible.For Nest cameras, migrate recordings to local storage using Frigate NVR (open-source, self-hosted). For Ring, examine your data sharing settings and disable law enforcement data sharing if that’s a concern for you (Settings → Control Center → Video Requests). The fewer copies of your footage sitting on vendor servers, the smaller your breach exposure.
- Enable multi-factor authentication on every account that supports it.MFA defeats the majority of credential-stuffing attacks, which account for a significant share of smart home account compromises. Use Authy or a hardware key like YubiKey for critical accounts. Do not use SMS-based MFA for accounts protecting cameras or door locks — SIM-swapping attacks make SMS MFA much weaker than app-based alternatives.
- Inventory your devices with Fing — including ones you’ve forgotten about.The Fing app scans your network and lists every connected device with manufacturer, IP, and MAC address. You will almost certainly find devices you forgot existed — an old security camera from three years ago, a smart TV that got connected once. Devices you’ve forgotten about are devices you haven’t updated or secured. Decommission anything that doesn’t need to be connected.
- Consider switching to Home Assistant for privacy-focused local control.Home Assistant is an open-source home automation platform that runs entirely locally on a dedicated device (like a Raspberry Pi or Intel NUC). No cloud dependency, no vendor data collection, full control. It integrates with nearly all major smart home ecosystems via Zigbee, Z-Wave, and Matter. It has a learning curve — but for privacy-conscious users, it’s the gold standard. Start here.
The Tools That Actually Work
Your Smart Home Security Checklist
- Full device inventory completed (every connected device documented)?
- All default passwords changed to unique strong credentials?
- IoT devices isolated on dedicated VLAN or guest network?
- VPN active at router level — not just on laptop?
- Firmware updates run across all devices?
- App permissions reviewed and excessive ones revoked?
- MFA enabled on all accounts (app-based, not SMS)?
- Pi-hole or equivalent DNS filtering installed?
- Camera storage migrated to local or at minimum reviewed?
- Quarterly review date calendared?
A household in the Pacific Northwest discovered their older security camera had been compromised after noticing unusual outbound traffic in their router logs. The camera, purchased in 2021, was still running 2022 firmware and had never had its password changed from the factory default.
The investigation (using Fing and Pi-hole logs) revealed the camera had been making regular connections to an unfamiliar IP address in Eastern Europe for at least three months. No footage appeared to have been exfiltrated — but the device had been enrolled in what appeared to be a botnet.
Their response: full network inventory (turned up two more forgotten devices), VLAN segmentation, Home Assistant migration for all smart home control, Frigate NVR for all cameras. Six months later, their Pi-hole dashboard showed zero suspicious outbound connections. Total cost: roughly $80 in hardware and two weekends of setup time.
Key lesson: The threat wasn’t exotic. The defense wasn’t complicated. It just required someone to look.
The Six Mistakes Even Tech-Savvy People Make
1. Trusting the brand name
Apple HomeKit genuinely does emphasize local processing and strict device certification — it’s meaningfully better on privacy than most alternatives. But “better” isn’t “immune.” Ring, owned by Amazon, had documented issues with law enforcement data sharing that continued even after policy changes. Vet devices specifically, not brands generally. The CVE database lists disclosed vulnerabilities per device and version — worth checking before you buy something that will live in your bedroom for five years.
2. Setting up the guest network and calling it done
Guest networks are a start. They’re not the same as a properly configured VLAN with firewall rules preventing IoT-to-trusted-network communication. On most consumer routers, devices on the “guest” network can still communicate with each other. Proper segmentation requires firewall rules blocking inter-VLAN traffic. If your router doesn’t support this, it might be time for a pfSense box.
3. Ignoring the “smart” TV
The NETGEAR/Bitdefender report found the greatest cybersecurity vulnerabilities in smart homes lie in entertainment and surveillance equipment — not in the obvious targets like smart locks. Smart TVs use Automatic Content Recognition (ACR) to identify everything displayed on screen and log it. Most manufacturers enable this by default. Disable it. On Samsung: Settings → Support → Terms & Privacy → Viewing Information Services. On LG: Settings → General → Live Plus.
4. Doing a one-time setup and never revisiting
A device that was secure in 2023 might not be secure today. New vulnerabilities are discovered constantly. Set a quarterly calendar reminder to run firmware checks, review device inventory, and check whether any connected apps have changed their terms of service or data-sharing practices. This stuff doesn’t stay fixed on its own.
5. Over-relying on cloud-dependent systems
If your smart home automation depends entirely on third-party cloud servers, you’re trusting that company to stay solvent, maintain their service, and keep their security infrastructure current. Companies shut down, get acquired, and change their privacy policies. Local processing via Home Assistant eliminates most of this dependency. It’s a bigger upfront investment. Worth it.
6. Not telling your household about the threat model
The most technically sound network setup can be bypassed by a family member who connects a new cheap IoT device directly to the main network and doesn’t mention it. Hold a 15-minute family conversation about why IoT devices go on the separate network, why new devices need to be discussed before connecting, and what to do if something seems wrong (weird noises from a camera, devices behaving strangely). Human-layer security is not optional.
The Horizon: 2026–2027 Smart Home Security Trends
A few things are worth watching, because they’ll change the landscape meaningfully over the next 18 months.
The Matter protocol is genuinely improving things. Matter is an interoperability standard now backed by over 200 vendors — Google, Apple, Amazon, Samsung all participate. Beyond making devices talk to each other across ecosystems, Matter builds security in at the protocol level rather than leaving it to individual manufacturers. Devices certified for Matter undergo more rigorous security review than legacy IoT devices. This is slow progress, but real progress.
Edge AI processing is shifting data flows. The privacy problem with most smart home AI is that it requires sending audio and video to vendor servers for processing. Edge AI — running inference locally on the device or a local hub — eliminates that requirement. Apple HomeKit has led on this; other ecosystems are catching up. By 2027, most premium smart home devices will process AI locally by default, which is a meaningful privacy improvement.
Quantum-resistant encryption is coming but isn’t here yet. Current encryption standards are potentially vulnerable to quantum computing attacks. NIST finalized its first post-quantum cryptographic standards in 2024. Expect device manufacturers to start implementing them in 2026–2027 firmware updates. For now, this is background awareness rather than an immediate action item.
Regulatory pressure is increasing. The EU’s GDPR extensions and “privacy by design” requirements for IoT are already shaping product decisions for manufacturers selling in Europe. U.S. momentum is slower but building. The practical implication: more manufacturers are embedding better security defaults because they have to, not just because they want to. That’s genuinely good for consumers.
Questions People Actually Ask
The Bottom Line
Here’s what I want to leave you with.
The threat is real. 13.6 billion attacks on consumer IoT devices in 2025 is not a number I’m inflating — it comes from analysis of 6.1 million actual households. 29 attack attempts per day on your home network. One in three data breaches involving an IoT device. These aren’t scare statistics designed to sell something. They’re the measured output of a security landscape that has expanded enormously as adoption has accelerated.
But the fixes are not complicated. The vast majority of successful smart home attacks exploit the same handful of failures: default passwords, unpatched firmware, flat networks, excessive permissions. Fix those, and you’ve dramatically reduced your risk. Not to zero — nothing is zero — but to a level where you’re no longer the easy target that botnets and opportunistic attackers are looking for.
The more ambitious path — Home Assistant, local processing, full network segmentation — is genuinely achievable for anyone comfortable with technology, and it gets you close to genuine privacy. It’s a weekend project, not a career. The guides exist. The community is large. The tools are free.
The smart home is not going away. The question is whether you’re its user or its product.
Ready to start? The full smart home security guide on ForbiddenAI walks through the Home Assistant migration in detail. For network architecture specifically, see our VLAN segmentation for beginners guide. Questions? Leave them in the comments — I read everything.
Hidden AI Conspiracies in Social Media Algorithms 2026: What Small Businesses Need to Know
7 Emerging AI Threats in 2025 That Started as Conspiracies: Protect Your Business and Cut Risks by 50%
You’re Using AI Every Day Without Realizing It(25 Honest Examples)
These AI Tools Save You Hours Every Day(And You’re Not Using Them Yet)
Real-Life AI That Saves Lives in 2025 (and Risks Them)
AI Conspiracy Theories That Sound Too Real: A 2025 Data Analysis
