root@bigbadman:~# cat iot-ev.html
iot + ev

> Your Connected Life Is Your Weakest Link

You secured your laptop with GrapheneOS on your phone. You use Signal for messaging. You're not clicking suspicious links. But what about your thermostat? Your baby monitor? Your smart TV? Your electric car charging at night? These aren't just appliances anymore β€” they're computers connected to the internet, running outdated firmware, vulnerable to remote exploitation, and often incapable of receiving security updates.

The Internet of Things (IoT) has exploded. By 2025, there were over 15 billion connected devices worldwide. That's two devices for every person on Earth. And almost all of them are insecure. Manufacturers prioritize cost and features over security. Security updates are optional, sometimes non-existent. Default passwords are ubiquitous. Encryption is an afterthought.

Meanwhile, electric vehicles are essentially computers on wheels. Teslas, BYDs, Volkswagens, BMWs β€” they all have internet connectivity, over-the-air updates, mobile apps that control the car, and dozens of electronic control units (ECUs) talking to each other. A 2023 study found that the average modern car has more than 100 million lines of code. More than a fighter jet. More than Windows. And many of these systems are remotely exploitable.

This page is about the invisible attack surface in your home and your garage. The things you don't think about until someone uses them to own your entire network.


🌐> Why IoT Security Is Broken

The fundamental problem isn't technical β€” it's economic. Security costs money. It slows development. It requires ongoing maintenance. In the race to get cheap gadgets to market, security is the first casualty.

  • No patching lifecycle β€” 60% of IoT devices receive zero security updates after purchase. When a vulnerability is discovered, it stays unpatched forever.
  • Hardcoded credentials β€” thousands of cameras, routers, and printers shipped with the same admin password. Mirai botnet alone infected 600,000 devices in 2016 using hardcoded credentials.
  • Unencrypted communications β€” many devices communicate with their cloud servers in plaintext. Credentials, telemetry, control commands β€” all sniffable on the local network or ISP level.
  • Insecure APIs β€” mobile apps that talk to devices often have weak authentication. API endpoints that allow unauthorized control of the device.
  • Supply chain compromise β€” firmware updated through third-party CDNs or third-party hosting. Compromise one vendor and you own all their customers' devices.
  • Local network exposure β€” IoT devices are typically on the same LAN as your laptops and phones. Once an attacker owns the smart bulb, they own your home network.
Average IoT Device Risk Level

The Mirai botnet proved how dangerous insecure IoT can be. In October 2016, Dyn's DNS infrastructure was taken down by a DDoS attack launched from 100,000 compromised IoT devices. Twitter, Netflix, Reddit, GitHub, Spotify β€” all offline for hours. The attackers didn't hack any major company. They hacked webcams, DVRs, and routers β€” devices whose owners never knew they were participating in a cyberattack.


🏠> Smart Home Devices: Ranked by Danger

Not all IoT is created equal. Some devices are relatively safe. Others are nightmares waiting to happen. Here's the ranking from least to most dangerous.

Smart Speakers
// Amazon Echo, Google Nest Home // Risk: LOW-MEDIUM // Network Exposure: Low

Smart speakers like Echo and Nest Home are listening constantly for their wake words. They send audio snippets to Amazon/Google servers for processing. Privacy concerns are real β€” but the attack surface is limited. These devices generally can't do much beyond controlling other smart home devices, playing music, and answering questions.

Risks

  • Voice recording leaks β€” employees at Amazon and Google review voice clips for "training"
  • Phishing via voice β€” researchers demonstrated sending malicious URLs through smart speakers
  • Wake word spoofing β€” ultrasonic signals that trigger devices without humans knowing
  • Network snooping β€” if on same LAN as other devices, could potentially probe for vulnerabilities

+ Attack Vectors

  • Compromised Amazon/Google cloud account
  • Malicious third-party skill/action
  • Wi-Fi network sniffing
  • Physical access reset

- Real Impact

  • Audio surveillance
  • Smart home control via voice injection
  • Data exfiltration to cloud
  • Low physical risk
Risk Assessment

Mitigation: Place speakers away from sensitive conversations. Disable voice history in settings. Put them on a separate VLAN if possible. Use mute button when discussing private matters.

Smart Cameras & Baby Monitors
// Ring, Nest Cam, TP-Link Tapo // Risk: HIGH // Network Exposure: Medium-High

Video cameras are everywhere now β€” doorbell cams, indoor monitors, baby cameras, pet cams. They stream video 24/7 to the cloud. Many have terrible security. In 2023, hundreds of thousands of Ring accounts were compromised through credential stuffing (password reuse from other breaches).

But here's the real horror story: baby monitors. In 2024, a researcher demonstrated taking control of a popular brand's camera remotely β€” able to speak through the speaker, pan/tilt the camera, and view the feed. Parents reported strangers talking to their children through the baby monitor. Not a one-off incident. Multiple brands affected.

Risks

  • Unauthorized viewing β€” hackers watching you through your own cameras. Happens more than you think.
  • Recording leakage β€” footage accessed by unauthorized parties, leaked online, or sold.
  • Two-factor authentication bypass β€” some brands implement 2FA poorly (SMS-based, or none at all).
  • Cloud API compromise β€” if the manufacturer's cloud is hacked, all customer feeds could be exposed.
  • Physical tampering β€” if the device can be physically accessed, it can be reset and reconfigured.

+ Attack Vectors

  • Credential stuffing (password reuse)
  • Weak or no 2FA
  • Cloud API vulnerabilities
  • Firmware exploits
  • Default credentials never changed
  • Man-in-the-middle on unencrypted streams

- Real Impact

  • Home invasion reconnaissance
  • Stalking/harassment
  • Child endangerment
  • Extortion (threatening to leak footage)
  • Privacy violation at scale
Risk Assessment

// REAL INCIDENT: In 2024, Ring revealed that 60,000 accounts were compromised between 2018-2020 due to credential stuffing. Hackers gained access to live video feeds and used the two-way audio to threaten residents. Ring responded by implementing mandatory 2FA, but the damage was done.

Mitigation: Choose cameras with end-to-end encryption. Enable 2FA (app-based, not SMS). Regularly change passwords. Review active sessions. Consider local-only cameras (no cloud dependency).

Smart Thermostats & Climate Control
// Nest Learning Thermostat, Ecobee // Risk: MEDIUM // Network Exposure: Low

Thermostats seem benign β€” they control your heating and cooling. But they're always connected, have remote access via apps, and if compromised, can be used for denial-of-service attacks against your HVAC system (turning heat to max in summer or off in winter).

Risks

  • Comfort sabotage β€” extreme temperatures while you're away
  • Energy cost inflation β€” constantly adjusting temperature up/down
  • Lifestyle inference β€” when thermostat is active reveals occupancy patterns
  • Lateral movement pivot point β€” thermostat compromises network, then attacker moves to cameras/laptops
Risk Assessment

Mitigation: Put thermostats on guest Wi-Fi or separate VLAN. Use strong unique passwords. Monitor for unusual activity logs.

Smart Plugs & Light Bulbs
// TP-Link Kasa, Philips Hue, Wyze // Risk: LOW-MEDIUM // Network Exposure: Low

These are the safest IoT devices. They don't collect much personal data, can't spy on you, and have limited functionality. But they're still computers on your network. If compromised, they could be part of a botnet or used as a foothold to explore your network.

Risks

  • Botnet participation (DDoS attacks)
  • Network scanning pivot point
  • Power manipulation β€” turning off equipment (minimal impact usually)
  • Pattern inference β€” light patterns reveal occupancy
Risk Assessment

Mitigation: Still worth putting on isolated VLAN. Buy from reputable brands that provide regular firmware updates. Disable cloud access if local control suffices.

Smart Locks & Doorbell Cameras
// August Lock, Yale Assure, Ring Video Doorbell // Risk: CRITICAL // Network Exposure: HIGH

This is where IoT meets physical security. A hacked smart lock means someone can unlock your front door from anywhere on Earth. A hacked doorbell camera means intruders know exactly when you're home, where you keep valuables, and whether your alarm is armed.

Historical Precedent

In 2022, Yale's smart locks were found to have a critical vulnerability β€” attackers could brute-force the PIN pad remotely via the wireless protocol. A similar flaw was found in August locks in 2021. The companies issued patches, but thousands of locks remained vulnerable for months because users don't update firmware religiously.

Risks

  • Remote unlock β€” attacker gains control and unlocks door. Home invasion made trivial.
  • Unlock timing β€” learning your schedule by monitoring when doors are unlocked (you leaving, coming home).
  • Access logs deletion β€” intruders covering tracks by clearing digital audit logs.
  • Guest code abuse β€” generating temporary codes for unauthorized people.
  • Firmware downgrade attacks β€” forcing device to use older, vulnerable firmware version.
  • Bluetooth relay attacks β€” intercepting and replaying Bluetooth signals from your phone to the lock.

+ Attack Vectors

  • Cloud API vulnerabilities
  • Weak authentication (no 2FA)
  • Bluetooth/Wi-Fi protocol exploits
  • Firmware reverse engineering
  • Physical tampering (if accessible)
  • Social engineering support staff

- Real Impact

  • Home invasion
  • Theft
  • Assault/Kidnapping
  • Total loss of physical security
  • Insurance fraud implications
Risk Assessment

// REALITY CHECK: Despite the risks, smart locks can be secure IF chosen carefully. Look for: end-to-end encryption, mandatory 2FA, local fallback (physical key), offline mode, independent security audits, and active vulnerability disclosure program. Avoid cheap, no-name brands. Stick with companies that take security seriously (or have been forced to by reputation).

Mitigation: Never use smart locks as sole security β€” always have traditional deadbolt backup. Enable 2FA. Check for firmware updates monthly. Physical inspection for tampering. Disable cloud access if local hub suffices.

Connected TVs
// Samsung Tizen, LG WebOS, Roku // Risk: MEDIUM // Network Exposure: Medium

Modern TVs are basically Android boxes with screens. They have microphones (voice control), cameras (on some models for gestures), Wi-Fi, and run apps. They also collect massive amounts of viewing data, often shared with third parties by default.

Risks

  • Audio/video spying β€” Samsung admitted in 2017 privacy policy that their smart TVs could "listen" to conversations. Microphone access can be abused.
  • Advertising tracking β€” TV watches what you watch, builds profile, sells to advertisers. Often opt-out only.
  • App vulnerabilities β€” Netflix, Hulu, YouTube apps can potentially be compromised, serving malicious content.
  • Lateral network access β€” once TV is owned, attacker can scan for other devices on same LAN.
  • DRM circumvention β€” if TV's content protection is cracked, pirated content can be distributed.
Risk Assessment

Mitigation: Disconnect microphone/camera in settings. Review privacy settings and opt-out of data sharing. Put TV on separate VLAN. Don't log into streaming accounts that store credit cards.


πŸš—> Electric Vehicles: Computers With Wheels

An electric vehicle (EV) is fundamentally different from a gasoline car. Every gasoline car has electronics β€” engine control unit (ECU), transmission, ABS, airbags. But EVs have more computers, more sensors, and crucially: internet connectivity.

Tesla Model S has ~100 ECUs. Modern EVs stream data to manufacturers continuously: location, speed, battery health, braking patterns, climate settings, even cabin audio (for "customer service"). Tesla's CEO claimed in 2021 that their cars generate 25 GB of data per day. Multiply that by 4 million Teslas β€” that's petabytes of telemetry.

And every connection is a potential attack vector.

Tesla
// Connectivity: 4G/LTE, Wi-Fi // Apps: Remote control, tracking // Security Research: Extensive

Tesla is the most researched EV from a security perspective. Which means we know exactly what vulnerabilities exist β€” because security researchers have published them.

Known Vulnerabilities (Publicly Disclosed)

Year Vulnerability Impact Status
2016 Remote door unlock via web API Vehicle access without physical key Patched
2018 Chrome browser exploit chain Full car control from parked state Patched
2020 Keyless entry relay attack Relay signal from phone to unlock Mitigated
2021 Sentry Mode camera exploitation Camera feeds accessible remotely Patched
2022 Bluetooth authentication bypass Unlock/start without app authorization Patched
2023 Charging station exploit Take control via Supercharger communication Under investigation

Theoretical Attack Scenarios

  • Remote immobilization β€” attacker stops car mid-drive. Safety hazard. Possible if brake/accelerometer data can be manipulated.
  • Steering control β€” steering is electronically controlled. Remote takeover theoretically possible with enough privileges.
  • Brake disabling β€” most EVs use regenerative braking + hydraulic backup. Both could be targeted.
  • Data exfiltration β€” location history, driving patterns, home/work addresses, frequent visitors.
  • Sabotage β€” thermal runaway induced by manipulating battery management system. Rare but possible.

// GOOD NEWS: Tesla takes security seriously (relatively). They have a bug bounty program ($$$ for responsible disclosure). Over-the-air (OTA) patches deploy rapidly. No known successful remote hijackings of moving vehicles in the wild. But "no known incidents" doesn't mean "impossible."

Overall Risk Assessment

Mitigation: Enable PIN to Drive. Disable remote features when not needed (keep phone nearby instead of relying on app). Park in well-lit areas with surveillance. Use Sentry Mode (even if it has had vulnerabilities). Keep software updated. Disable cellular when parked long-term (some models support this).

Chinese EVs (BYD, NIO, XPeng, Polestar)
// Growing global market share // Data sovereignty concerns // Security audits: Limited

Chinese-made EVs are flooding Western markets. BYD just surpassed Tesla in global sales in 2024. This presents unique security challenges:

Concerns

  • Data routing β€” Chinese regulations require data from Chinese companies to be stored in China. Your driving data may legally reside in Beijing servers.
  • Government access β€” Chinese National Intelligence Law (2017) requires Chinese organizations to assist state intelligence work. Theoretically, data could be compelled.
  • Supply chain opacity β€” hard to verify what components are in these vehicles and their origins. Chip supply chains are notoriously opaque.
  • Backdoor allegations β€” multiple Chinese tech products have been found with undocumented remote access capabilities (XZ backdoor, Huawei concerns). Cars are no exception.
  • Security research bans β€” Chinese companies actively sue/restrain security researchers from publishing vulnerabilities.

// GEOPOLITICAL CONTEXT: In 2024, EU began investigating Chinese EVs for security risks. US already banned Chinese software in cars from government fleets. This isn't just paranoia β€” governments take these threats seriously because vehicles are now critical infrastructure.

Overall Risk Assessment

Mitigation: If you must use these vehicles, disconnect cellular entirely. Use offline mode. Physically block microphones/cameras when not needed. Don't link personal accounts (email, cloud storage) to the vehicle.

Traditional Automakers Going Electric (Ford, VW, GM)
// Ford Mustang Mach-E, VW ID.4, Chevy Bolt // Security posture: Varies

Legacy automakers are scrambling to electrify. Their security maturity varies wildly. Volkswagen has invested heavily in cybersecurity (they employ dedicated security teams). GM had a catastrophic recall in 2014 when a software bug disabled ignition switches β€” causing deaths. Ford has mixed history.

Key Differences From Tesla

  • Less telemetry β€” generally, legacy makers collect less data than Tesla (though still substantial).
  • Slower OTA updates β€” patch cycles are months, not days. Vulnerabilities stay open longer.
  • Weaker security programs β€” few have robust bug bounties or coordinated disclosure.
  • More attack surface β€” often retain legacy CAN bus architecture alongside new digital systems. Old systems weren't designed for internet connectivity.
Overall Risk Assessment

Mitigation: Same as Tesla β€” PIN to start if available, disable unnecessary connectivity, keep software updated, park securely. Research specific model vulnerabilities before purchase (security researchers often publish CVEs for popular vehicles).


πŸ”‹> Charging Infrastructure Risks

Your car is only half the equation. Where you charge it matters too. Public charging stations are connected computers running Linux (often), communicating with your car via proprietary protocols, and billing through payment processors.

Attack Vectors at Charging Stations

  • Plug-and-charge exploitation β€” ISO 15118 protocol allows automatic authentication. Researchers demonstrated man-in-the-middle attacks stealing credentials.
  • Firmware updates via charger β€” some chargers push firmware to vehicles. Compromised charger = compromised car.
  • Payment data theft β€” skimming devices, SQL injection in backend databases, credential harvesting at station interfaces.
  • Grid sabotage β€” charging infrastructure connected to electrical grid. Malicious actors could overload transformers, cause localized blackouts.
  • Social engineering β€” fake charging apps that steal account credentials. Fake QR codes on broken chargers redirecting to phishing sites.

// RESEARCH UPDATE: In 2023, security firm PenTest Partners published a paper showing they could remotely control a Tesla through its Supercharger connection. Car was stationary, parked at charger. They achieved this through the PLC (power line communication) channel. Patch released within weeks, but demonstrates the attack surface.

Mitigation: Use trusted networks (home charging ideal). Avoid public chargers for critical travel (battery anxiety makes you desperate). Check charger integrity before plugging in (look for tamper signs). Use physical adapter for payment instead of storing card digitally.


πŸ›‘οΈ> Hardening Your IoT Ecosystem

You can't eliminate all risk. But you can raise it to the point where casual attackers move on and targeted attacks become expensive. Here's the practical checklist.

Network Segmentation

  • VLAN isolation β€” Put all IoT devices on separate VLAN from main devices (laptops, phones). Use router/firewall rules to block IoT from initiating connections to main network.
  • Guest network β€” if VLAN is too advanced, at minimum use guest Wi-Fi for IoT. Guest networks typically can't access main subnet.
  • Router choice matters β€” pfSense, OPNsense, Ubiquiti UniFi. Cheap ISP routers lack granular firewall controls.

Device Selection Criteria

  • Check update history β€” visit manufacturer's security advisories. No updates in 2+ years? Skip it.
  • Look for security certifications β€” UL certification, ETSI EN 303 645 (EU IoT security standard), ANSI/UL 2900.
  • Prefer local control β€” devices that work offline via local API are safer than cloud-dependent ones.
  • Avoid no-name brands β€” Amazon Basics devices, AliExpress specials, random Kickstarter campaigns. They vanish when vulnerabilities are found.

Authentication Hygiene

  • Change default credentials β€” yes, really. Write them down in password manager if needed.
  • Mandatory 2FA β€” every device that supports it. Prefer authenticator app over SMS.
  • Unique passwords per device β€” don't reuse your main email password on your smart bulb account.
  • Disable unused features β€” remote access, cloud backup, voice assistant integration β€” turn them off if you don't use them.

Monitoring

  • Network traffic analysis β€” use tools like Wireshark, ntopng, or firewall logs to spot anomalies. IoT devices shouldn't be making thousands of outbound connections.
  • Unusual behavior alerts β€” if your fridge suddenly starts downloading 10GB of data, investigate immediately.
  • Log review β€” check device admin consoles monthly. Look for unrecognized login locations or access times.

Physical Security

  • Lock routers/firewalls β€” keep them in locked cabinets. Physical access = game over for most security measures.
  • Secure cameras β€” don't point them at entry points where passwords might be entered. Cover them when not needed.
  • Factory reset before disposal β€” wipe everything. Some devices have hidden partitions that survive normal resets.

⚠️> The EV Purchase Decision

Should you buy an electric vehicle given all these risks? Here's a framework to evaluate based on your threat model.

User Profile Recommendation Rationale
General consumer Safe to proceed Risks are theoretical. Millions of EVs on road. No widespread hacks observed. Benefits outweigh risks.
Corporate executive Caution advised High-value target. Consider Chinese-brand restrictions. Disable connectivity features.
Journalist/activist in hostile regime Avoid entirely Location tracking, remote immobilization, data access by authoritarian governments are real threats.
Security researcher Research first Read CVEs for specific models. Bug bounty programs indicate vendor responsiveness.
Privacy maximalist Delay purchase Wait for open-source EV projects (Polestar OS open source initiatives, community mods).
Government employee Check compliance Many countries restrict foreign-made EVs for government use. Follow procurement guidelines.

For most people, EVs represent a net positive for security and sustainability. The environmental benefits are real. The convenience factors (remote preconditioning, over-the-air updates fixing bugs) are tangible. The attack vectors exist but remain largely theoretical in practice.

The exception: if you're a high-value target (journalist, activist, corporate insider, government official, whistleblower), the risk calculus changes dramatically. Your threat model includes nation-state adversaries. Nation-states have budgets for vehicle exploitation. In that context, every connected device is a liability.


> The Bottom Line

IoT and electric vehicles represent convenience traded for security. You gain remote access, automation, efficiency. You lose control over your attack surface. That's not necessarily a bad trade β€” it depends on your priorities.

But the trade-off must be informed. Most people buy smart thermostats and EVs without reading the privacy policy, checking the security track record, or understanding the network implications. That's negligence disguised as trust.

Here's the truth no marketing department will admit:

  • Every connected device collects data β€” even if "we don't sell your data," you're still being profiled internally.
  • Security updates are rarely guaranteed β€” unless explicitly stated (GrapheneOS promises 5 years, Tesla does OTA regularly), assume no patches after 2 years.
  • Vulnerabilities are inevitable β€” every complex system has bugs. The question is whether the vendor responds responsibly when flaws are found.
  • Your network is only as secure as its weakest link β€” a compromised smart plug can be the gateway to your entire LAN.
  • Physical and digital security converge β€” hacked smart locks enable burglaries. Hacked cars enable kidnappings. The stakes aren't just data loss anymore.

// FINAL WORD: You don't have to go Luddite. You don't have to live without modern conveniences. But you do need to think critically about what you bring into your home and onto your road. Security is a process, not a product. Question everything. Verify assumptions. Update relentlessly. And remember: if you're not paying for the product, YOU'RE the product. Always.

// every plug is a portal. every sensor is a spy //