This underscores the principle that IoT devices should not be allowed to communicate over the public Internet. Pretty much all cheap, Chinese-made hardware of this kind has intentional or unintentional security holes waiting to be exploited.
This underscores the principle that IoT devices should not be allowed to communicate over the public Internet. Pretty much all cheap, Chinese-made hardware of this kind has intentional or unintentional security holes waiting to be exploited.
> Pretty much all cheap, Chinese-made hardware of this kind has intentional or unintentional security holes waiting to be exploited.
Why single out bad Chinese coding? Bad US IoT coding has a longer history.
There’s bad, and then there’s egregious.
Plenty of US companies have egregious vulnerabilities
Like iRobot recording people on the toilet and uploading the videos?
Yeah, I agree - at least Chinese Roborock gives very granular controls for privacy.
There’s egregious and there’s malicious.
Yeah, we’ve seen that as well.
All of there IoT devices will be slop coded soon, and I wonder whether that will be an improvement or not. I bet that security will be better.
> I bet that security will be better.
Not doxing myself, but... Company with a known name vibecoded a dashboard with Claude. Which also hardcoded a password into the client-side of the dashboard, which I caught.
I reckon security will be about the same.
> Not doxing myself
Seems irrelevant to the comment to add this, James. It just screams to do it a-la Streisand effect..
When I'm reading reviews of plans created by an agent especially on security boundaries it's suggesting huge matrixes to test even the very obscure situations, but then I'm also reading things like this and I just don't understand. Are we even using the same tools?
A large part is also how much you read back what the model writes. The good models generally write quite secure code, but they also often implement temporary solutions that they tell you to fix later.
For example, if secret storage methods aren't specified in the prompts, a model might decide to be clever and implement a generic secret access interface, with a default implementation that hardcodes everything. It will probably tell you that this is not production ready and you should write or specify your preferred secret storage implementation, but if you don't read or understand that, you'll just leave it as is and push to prod.
Management think models mean juniors can do senior work. Juniors don't know the footguns. Juniors can't read the code that the system outputs. Models get overwhelmed in any decent sized codebase.
Why would you be surprised there are failures?
[flagged]
Prod bugs up 260% since AI approval? Yeah, I'm gonna disagree.
BAs using AI make worse designs, management makes seniors redundant making code worse, and then QA is also laid off to make room for agentic testing. The results are not a surprise in anyway.
Refusal to consider evidence to the contrary is certainly symptomatic of something, and it isn't a well-reasoned argument. "Us" sounds like the corporate leadership I work with that have actively pushed teams to incur mountains to tech debt. I was walking over org metrics this week and sure enough, there's a rising tide of unplanned work for bugs, performance problems, and security snafus introduced during an aggressive push into GenAI with pressure to "just make agents do it."
All of which was, of course, predicted by management, reported upwards, and ignored.
1. It depends on model and tokens spent 2. Models talk the talk but not always walk the walk
Tools are already preventing IoT companies from doing a ton of things they do, by default. It's a problem of the process, churn, and culture, not tools. I don't have any doubts that if given a coding agent that cares more than they do they'd still force it to hardcode a password or something because they feel like it's more convenient. Nobody cares there.
> All of there IoT devices will be slop coded soon
Soon?
I've already seen multiple of TP-Link's firmware engineers leave their LLM history public and indexed by search engines.
It's quite obviously them as well.
How did you find that?
>Why single out bad Chinese coding?
You know precisely why anyone would single out China here. They are egregiously bad, and you know it, and everyone else knows. No amount of "what about this other bad thing that's also bad" could possibly allow any normal person to escape this conclusion.
Better to buy devices that can work without internet and just blacklist them at the router level. Price or origin is not a good metric to ensure no leaks.
Which is very hard, because most of these devices don't work locally. They communicate via the cloud to your app.
In most cases companies don't want to give you Matter or HomeKit, because it means they cannot sell you more through their app.
Wyze has ads everytime you open it. So does Honeywell. Hell, even the internet-loved Ecobee has a banner that shifts everything down most of the time that you open the app. And for that last one, you _have_ to use their app to control the fan, as they don't expose separate fan controls over HomeKit...
> Which is very hard, because most of these devices don't work locally.
Then don't buy those devices. All of my home devices are either Zigbee (local-only) or were bought specifically because they can be reflashed with open firmware to liberate them from the cloud.
> Hell, even the internet-loved Ecobee has a banner that shifts everything down most of the time that you open the app.
I don't want such apps. The only app I need is Home Assistant, and installing proprietary software on my phone (running GrapheneOS) is out of the question.
> Chinese-made hardware
Honestly, I'd rather it leak my GPS to the Chinese government than the US government. They don't have jurisdiction over me anyway.
> should not be allowed to communicate over the public Internet
It would be a no-go for non-techies. One of the biggest draws to IoT devices for "average Joes" is being able to view and control them from remotely, and they aren't going to have the skills or know-how to set up a VPN correctly with dynamic DNS so that their phone can VPN into their home and then sideload/jailbreak their phone to load a custom app to control it. "It just works from anywhere" is a big sell for them.
> It would be a no-go for non-techies.
There are better solutions, like Apple’s HomeKit. I’m able to watch a camera that has no internet access because it passed through my Apple TV, which serves as a home hub. I didn’t have to set any of this up, it just works when you have the required hardware.
> I’m able to watch a camera that has no internet access because it passed through my Apple TV, which serves as a home hub.
How exactly does this prevent the same kind of issue for Apple devices? Aren't you just trusting that Apple handles your data better than TP-Link? Not saying they don't but routing through another device doesn't really add security on its own.
> Aren't you just trusting that Apple handles your data better than TP-Link?
I am, yes. Ultimately you’re going to need to trust some hardware, somewhere. No matter what you’re doing you have to trust that your home router doesn’t have an externally accessible SSH port with no password set.
Personally I trust Apple more than I trust TP Link with this stuff.
> Personally I trust Apple more than I trust TP Link with this stuff.
What if a TP-Link camera supported HomeKit Secure Video? You access the camera through Apple but all of these cameras are still directly connected to the internet, meaning you still need to trust the camera manufacturer.
> No matter what you’re doing you have to trust that your home router doesn’t have an externally accessible SSH port with no password set.
No, you don't have to trust. I build my own routers precisely because I don't.
Pretty sure that just means that we trust Linux/Openwrt and the chip vendors.
Linux and OpenWrt have an incredibly good security record compared to what ships on typical consumer devices.
Doesn’t it? I do trust the manufacturer of the thing that I keep within arms reach from the moment I get it, and which knows more about me than I do, to have better security than whatever the hell a TUYA is that controls the lights.
HomeKit will take care of the VPN/remote access part, sure, but your devices still need to communicate with the HomeKit device, and that's usually over Wi-Fi, which puts the devices on the public internet, and carries the same security risk.
There are various non-internet protocols for IoT devices, none of them good:
* Zigbee: Requires some technical understanding to set up, devices randomly disconnect for hours even when they are 2ft from the coordinator, all-around horrible experience for non-techies
* Non-standard Zigbee variants: even worse
* Matter-over-Thread: horrendously designed from a UX perspective. Easy-to-lose barcodes stuck on cards in the packaging, weird 12-letter codes, and your non-techie cannot understand what the hell Matter or Thread is. Pairing is an absolute nightmare.
> devices randomly disconnect for hours even when they are 2ft from the coordinator
I don't think that's normal. Like, to the point where I'm wondering if you have a bad opinion of the whole protocol because you got a faulty device.
Yes this. My experience with Zigbee is that it just works, but the setup and specific hardware probably matters. I use a SONOFF USB dongle through Zigbee2MQTT and Home Assistant.
> Zigbee
Requires no technical understanding. At least not more than e.g. a WIFI router.
> devices randomly disconnect for hours even when they are 2ft from the coordinator,
You present this like a fact. But it is at most an anecdote. I present you a different anecdote: I have ~30 zigbee devices, in two different houses (first a house with concrete floors and cellar and level 1..3) and now one old woodwork structure house with 2 floors. Nowhere did I had even half an hour of disconnection.
> all around-horrible
... excellent experience even for my ex-spouse, which is/was non-techie.
However, that you present Zigbee here at all is weird. Zigbee doesn't have any way to transport a camera stream. It's mean for low-powered battery devices. My temperature sensors got a 1500mAh AAA chargeable batteries and they lasts now for over one year. Note that I have sensors from ~ 15 different brands. Mostly battery powered sensors and mains power switchable plugs.
I also enjoy that these Zigbee devices are by design completely disconnected from any IP traffic. This, and their (intentional) low data rate make them almost impossible to misuse. E.g. as denial-of-service originators or amplifiers.
It's like you present WIFI as long-range thingy but actually you'd want LORA for that. I'm not assuming that knowing for what kind of usage a tech was designed as "needing technical understanding". After all, no one would claim "you need technical understanding" to know that you better use a truck instead of a Porsche Cayman to transport 50 cubic meters of sand.
> Nowhere did I had even half an hour of disconnection.
Well my garage door opener sensor has been disconnected for two 30 minute gaps today and my plant humidity sensors go offline for 2 weeks at a time.
So yeah, it's not ready for prime time.
> LORA
No, let's not even go there. Tech nerd protocol here that's an awkward middle ground that creates even more problems. Average Joes aren't going to set that crap up.
Are you using a phoscon coordinator? ConBee 2 has a lot of firmware problems in my experience.
There are also some devices which advertise ZigBee compatibility but the manufacturers don't seem to test them against coordinators other than their own (and ConBee 2 seems to have the most problems in this regard).
The protocol is complex, they all are, implementing it correctly isn't a given, but I think the issues people have are more often a factor of how long a protocol has been in use than any fundamental aspect of it.
As soon as cheap hardware manufacturers get on board you get this problem.
Quality hardware works fine with ZigBee. It's by no means perfect technology, if you want that, use copper wires, but it doesn't work as badly as you claim if you are not unlucky with coordinators and devices.
Is your zigbee running at 2.4GHz? Everything interferes with that.
Practically all consumer Zigbee devices only support 2.4GHz. You would need to go for ZWave for the sub-GHz range.
Also, it's not like 860-930 MHz (depending on the country) is without interference.
> My temperature sensors got a 1500mAh AAA chargeable batteries
You meant rechargeable? You seem to know more about Zigbee than about rechargeable batteries.
I dunno, through HomeKit, I've got a couple Zigbee networks (Hue and Ikea), a Z-Wave network (Home Assistant) and a Matter-over-Thread (Apple/Ikea) network, and they all seem pretty good?
Anything any government can access, motivated criminals can too.
Pissed off script kiddies have been confused as government plenty of times by unsuspecting victims.
I think no country can be trusted and therefore all IoT device communication (telemetry/updates) should go through government-controlled local servers in an unecrypted form. This way local government can enforce its policy over foreign manufacturers and has a "kill switch" if they misbehave.
Also, all communication and telemetry should be opt-in and turned off by default.
Government-controlled server would prevent foreign countries from collecting intelligence and pushing malicious updates.
> This underscores the principle that IoT devices should not be allowed to communicate over the public Internet.
TP-Link is a prominent maker of network hardware, including home and mesh routers.
And their network hardware is illegal to import into the USA because it's full of backdoors.
So, TP-Link isn't sold as consumer level devices in stores in the US?
It's had the Huawei treatment?
It's one of the primary networking electronics brands in Australia.
Yes it had the Huawei treatment
What are the details of that treatment?
As I write this, there are all kinds of TP-Link routers, mesh nodes, and cameras (oh my!) in stock at a nearby Wal-Mart, with pickup promised within a few hours from now.
TP-Link is absolute crap of network hardware. Not to mention “leftover debug code”: https://nvd.nist.gov/vuln/detail/CVE-2024-21827
Yea, where? :-)
This is why I like Zigbee and am very uncertain about Matter that basically allows (or maybe one day requires) internet access.
Consumers just don't care about security. It is what it is.
There is no reasonable way to assess security for the average consumer.
I agree. The market doesn't work well around this.
You can ask chatgpt.not a great way.
And when you ask it you the secure answers cost 3x more. And than require an installar. And some(Google) require a monthly subscription.
And even about the good systems, the chat recommends, since there's no mathematical guarantee for security, that you "Switch them off or physically cover the lenses while you are home.".
Even if there’d be a way, there’s no culture of asking questions about how things work, especially outside the single “happy” path.
Because there are too many things. We've built society on the notion that you don't have to know. Which of course unethical corporations try to exploit.
But if people knew how easy it was to use the camera they bought to spy on their family, then I bet many would care.
True, that's probably the biggest issue.
But the only solution here is very expensive marketing, so...
[flagged]
routers could solve this for consumers with a checkbox for "intranet only"
99% of consumers won't know how to setup a firewall but could handle a checkbox
only problem I have is I can't seem to punch a hole for time sync and it won't use my local intranet time server
Time server can be specified with DHCP, if you want. Options 4 [deprecated, but that doesn't matter] and 42.
No idea if that helps with your particular devices; they are, of course, free to ignore those fields.