Home · Personal · Tech & Digital Life · Smart home

PARTLY

As of 13 August 2026, AI can only partly secure your smart home network.

This still needs a person who signs their name to it.

Can you do it?

5 minutesto a draft.

30 minutesto something you’d act on.

Cost, all in£0

Skill neededchat-fluent

Who has to check ityou

What the alternative costsNo suitable commercial network-security product is listed in the supplied tool data.

If this goes wrong, you can lock yourself out, disconnect smart devices or leave a device exposed while believing it is protected.

What to actually do

  1. Hand it to a person

    The route this page recommends

    A person who owns the outcome does this end to end, worth it when the failure is dear.

  2. Use a tool built for this

    Second choice
  3. Do it yourself

    The distant third

    A chat interface, chat-fluent skill, and roughly 30 minutes until you can act on the result.

    How to actually do it

    1. Open your router manufacturer's support page and its administration app or web interface, then record the router model, firmware version, internet provider, Wi-Fi security mode and whether remote administration is enabled, without copying any passwords or keys.
    2. Open the manufacturer's support pages for your smart hubs, cameras and other connected devices, and make a list of each device, its app, its firmware version, whether it still receives updates, and whether you still use it.
    3. Paste the prompt and your non-sensitive list into a chatbot, answer its questions from the router and device screens, and reject any request for passwords, Wi-Fi keys or recovery codes.
    4. Use the returned checklist to change one setting at a time in the router app and device apps, starting with administrator access, Wi-Fi security, firmware and software updates, remote administration and unused devices.
    5. After each change, compare the router or app screen with the expected result in the answer, then test internet access, each smart hub, cameras, speakers and any automations that matter to you.
    6. Ask the chatbot to turn any missing, unclear or model-specific setting into a question for your router manufacturer's support team, and stop rather than applying advice that would expose a device or cannot be independently checked.

    Prompt

    Help me improve the security of my UK home smart-device network. Do not guess my router model, device settings, firmware status or internet provider. Ask me for the minimum information you need before giving model-specific instructions, and tell me where to find each item without asking for passwords, Wi-Fi keys, recovery codes, private IP addresses or other secrets. Start with safe checks covering the router administrator account, Wi-Fi encryption and password, firmware updates, remote administration, guest networking, UPnP, unused devices, default passwords, software updates, two-factor authentication and whether any device needs internet access. Separate advice for the router, phones and computers, cameras, speakers, plugs, hubs and other smart devices. Do not tell me to weaken encryption, expose a device directly to the internet, or change an advanced setting unless you explain the risk and give the exact way to undo it. For every recommendation, state what I should see in the relevant app or router page, how to test that existing devices still work, and what I should do if the option is missing. Mark anything that requires a network-security professional. Give me a short ordered checklist at the end. Never claim the network is secure just because the checklist is complete.

    Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.

What it gets wrong

  • AI cannot see your router configuration, connected devices or network traffic unless you provide accurate information from the relevant screens.
  • It cannot guarantee that a device has stopped receiving security updates or that its manufacturer has fixed a known vulnerability.
  • It cannot safely infer the effect of an unfamiliar router setting from its name alone, especially on an unusual ISP setup.
  • It cannot prove that every device is protected after a checklist, because a hidden device, weak account or outdated hub may remain.
  • It cannot take responsibility for lockouts, lost automations, privacy breaches or devices left exposed by a wrong change.

What caps this at PARTLY: context depth, verification cost and stakes of error.

How we scored this

Five axes, each scored nought to two by hand: ten means AI carries the task cleanly, and the thresholds that turn a total into YES, PARTLY or NO are published in the methodology. Each axis name links to its definition.

AxisScore (0–2)
Output2
Inputs1
Verification1
Liability1
Effort delta2
Total7 / 10

FAQ

Can ChatGPT secure my smart home network?
Partly. It can turn your router and device information into a practical security checklist, but it cannot inspect the network or guarantee that every device is protected. You must make the changes and check that they work.
What should I ask AI about my smart home security?
Ask it to check router administrator access, Wi-Fi encryption, firmware, remote administration, guest networking, unused devices, default passwords and updates. Tell it not to guess settings or ask for passwords, Wi-Fi keys or recovery codes.
Is it safe to give AI my router details?
Give it the router model, firmware version and non-sensitive setting names, but do not share passwords, Wi-Fi keys, recovery codes, public IP addresses or private IP addresses. Remove account names and serial numbers unless the manufacturer's support instructions specifically require them.
Can AI tell if my smart home network has been hacked?
Not reliably from a description or checklist. It can help you look for unfamiliar devices, changed settings and unusual account activity, but confirming an intrusion may require router logs, device evidence or a network-security professional.

Nearby answers

Assessed by gpt-5.6-luna (gpt-5.6-luna) on 2026-08-13, second-checked by an independent model. Wrong somewhere? Email [email protected] and it gets re-checked.

The newsletter

AI news, new answers and product picks, straight to your inbox.