To what extend do these "smart" products state how, when and what data they are collecting and selling?
Isn't it high time for legislation that at least makes it clear on the box what kind of sensors and integrations it has and how bad the product is privacy wise?
Also, shouldn't corporate spying disguised as automated content recognition and classification be limited or outright be banned?
And all this connected garbage should by law be able to function with the cord cut and without registration, and if an update negatively changes the advertised product characteristics the customer must be allowed to return it for full paid price even if it is 10 years old. Force the manufactures to make statements on the "box" about the features, the support duration and level, and the privacy characteristics.
They don't care because of several reasons.
1. They get paid not to
2. They are told not to.
Alot of this spying ends up producing valuable data for different agencies to purchase which otherwise they cannot legally obtain without a warrant. Sure some of it has some commercial value, but a lot of it is probably more valuable for domestic spy operations.
Remember guys, if they are reading it for wrongthink, they are reading it for industrial espionage.
Anthropic and openAI just announced they’re spinning out competitors to their biggest customers… right after ingesting all of their IP and contracts. Do the math.
(Or don’t, because they’ve been caught stealing math research)
> because they’ve been caught stealing math research
No. They haven't been caught stealing. Tristan Buckmaster publicly expressed surprise that the OpenAI model has used the niche approach he pursued (with AI assistance). 10,000 agents exploring various approaches make it much less surprising.
The "coincidence" defense never works for random individuals so I won't buy it when it comes companies worth billions which will lie, cheat, and deceive to make even more billions. They've been caught lying and stealing again and again. After so many years of this I think any trust put in them is either blissful ignorance or willful misdirection.
Claude is absurdly sensitive to anything that touches cyber, and threatens bans if it gets a whiff of a pentest going on. it wouldnt surprise me if openai had a similar approach.
Yesterday I joked that claude gave me a cyber stop for asking how to make the anime girl video for an exploit, but it was a thing that happened. Meanwhile im trying to convince people around me that having grok nudify people isnt "fine" just because grok says it is. Its wild where the lines a drawn.
I'm in the new CVP, and even Mythos gave me a cyber stop just for creating an MCP server that can remotely connect to my Steam Frame to run commands. It must've thought I was trying to create a centralized malware or something.
Not even sure why Mythos even still exists when it's identical to Fable in every way, including refusing to do stuff that looks like it could ever be repurposed into unapproved use cases.
Tastelessness aside, this seems perfectly rational to me. Unlike reverse engineering for the purpose of developing security exploits, "nudification" doesn't meaningfully expand the capabilities of malicious actors with access to free online porn and Photoshop.
That's one of the weirdest takes on here in a while.
I can develop security exploits myself just fine without the use of AI. Many people don't have the capability to use Photoshop and free online porn to deepfake someone else they might or might not personally know into a realistic video...
With Kimi K3.1 Pro, today WAS the day they actually caught up.
I’ve been saying this for the past few months - once the inflection point happens (such as today), it’s only a matter of 2 years when American closed models won’t be able to catch up with Chinese open models
The headline is misleading, but the "article" is only five screenshots of Twitter posts. Someone used AI to decrypt the Samsung TV's firmware and found always on listening, classification, lots of privacy nightmares. Then their (the hacker's, not Samsung's) OpenAI got suspended for "Cyber Abuse".
Right, but still: OpenAI provided a service and it's up to them to determine acceptable use. Samsung sells a $3000 (or actually $2999,95) TV and you get big brother watching you.
This is again will be very funny race to watch. Right now you can fairly do lots of reverse enginnering or "hacking" with off the shelf Chinese models, and they are very capable (GLM3 for example) which can be run locally if you stomach the hardware. Tools like rea are also becoming more popular. If openai continues to be pedantic, they will lose the technical base given how close is the gap getting.
The company I work for has a plan B to switch to alternative, self-hosted models if providers pull off strange moves, such as drastic pricing changes. We have already switched from one provider to another once, and we do not want to get burned again. We do not reverse engineer, just regular development.
Have you tried Qwen3.8-Flash-Next? I found it significantly better in anything I tested than GLM5.3-Flash, but the thing is you either use Alibaba cloud or you host it yourself (no lesser quant than unsloth 4.5 or mx4/nvfp4 and kv no more quantized than int8).
The Qwen3.8-Flash-Next model is ok for reverse engineering. It can definitely solve simple tasks, but it gets easily confused, and burns tokens like crazy.
(As an aside: I think representing tools for AI as virtual SQL tables is much better than traditional MCP servers, and that this approach deserves a lot more attention. Props to Elias Bachaalany for thinking out of the box and putting it together).
The results of the benchmark are below. It's sorted by cost, but I added notes about each AI model for myself. The notes try to talk about intelligence, speed, and cost. Intelligence is a subjective measure which I assign to each model by reading its thinking tokens and watching its tool calls. You can definitely tell which models are more intelligent, even on this small benchmarks, by their tool use, false starts, and how easily they fall for the simple misdirection in the crackme.
+----------------------------------+--------------------------------------------------------------------------------+-----------------+-----------------+-----------------+
| Model / route | Experience / outcome | Total tokens | Output tokens | Cost |
+----------------------------------+--------------------------------------------------------------------------------+-----------------+-----------------+-----------------+
| Xiaomi MiMo v2.6 Flash | Incredibly fast and intelligent; solved. | 225,291 | 3,281 | $0.005 (0.5c) |
| GLM 5.3 Flash / OpenRouter | Least intelligent group but cheapest; solved. | 340,006 | 11,205 | $0.01 |
| ChatGPT Luna Latest | Very fast, concise, token-efficient, intelligent, and cheap; solved. | 401,837 | 5,119 | $0.01 |
| Qwen 3.8 Flash | Initially confused and failed, then eventually solved. | ~418,000 | 16,676 | $0.02 |
| Qwen 3.8 27B | Fast; one wrong answer, then quickly self-corrected. | ~774,784 | 12,658 | $0.02 |
| DeepSeek Flash / direct API | Fast, cheap, and accurate; solved. | 369,334 | 9,089 | $0.03 |
| DeepSeek Flash / OpenRouter | Fast and intelligent; solved in 76 seconds. | 371,689 | 13,888 | $0.03 |
| Claude Haiku 5.5 | Fast and surprisingly intelligent, but overly thorough and token-heavy. | 662,346 | 26,244 | $0.03 |
| Thinking Machines Inkling | Low intelligence; good speed and relatively token-efficient; may loop. | N/A | N/A | $0.03 |
| Xiaomi MiMo v2.6 Pro | Overthought the task and wandered into tangents. | ~420,644 | 13,638 | $0.05 |
| GLM 5.3 / OpenRouter | Cheap and intelligent but behind DeepSeek; solved in about 2m 20s. | ~702,473* | 11,879 | $0.08 |
| GPT Sol Latest (Sol 6.1) | Slow, but highly intelligent and extremely token-efficient; solved in ~89s. | 294,957 | 1,406 | $0.15 |
| MiniMax M3 | Least intelligent comparison; many wrong turns and repeated small mistakes. | 4,546,552* | 116,804 | $0.20 |
| GPT Terra Latest | Very fast, very smart, and token-efficient; solved. | 229,888 | 2,342 | $0.20 |
| Mistral Large 4.0 | Medium speed; small mistakes but reached the solution. | 303,687 | 4,041 | $0.21 |
| Kimi K2.7 Code | Less intelligent, slow, expensive; not worth using. | 100,740 | 8,938 | $0.22 |
| Claude Opus 5.5 | Highest intelligence and token efficiency; pretty fast; cheaper than Sonnet. | 312,837 | 1,862 | $0.24 |
| Qwen 3.8 Max 0902 | Fast and intelligent, better than Kimi K3, but token-hungry. | 1,305,241 | 26,390 | $0.25 |
| Kimi K3 / OpenRouter | Faster and cheaper than direct K3, but still very expensive. | 750,064 | 5,922 | ~$0.32 |
| Meta Muse Spark Open | Intelligent and medium speed, but among the more expensive runs. | 872,850 | 10,284 | $0.35 |
| Gemini 4 Argon | Extremely fast and intelligent; strong value for capability. | 384,046 | 2,870 | $0.35 |
| Grok 4.7 | Very fast and intelligent, but expensive. | 655,648 | 4,217 | $0.46 |
| Meta Muse Glimmer 30B | Fast but low intelligence and token-inefficient; cancelled unsolved. | N/A | N/A | ~$0.50 wasted |
| Claude Sonnet 5.5 | Smart and fast but very token-inefficient and expensive. | 1,243,285* | 20,381 | $0.57 |
| Kimi K3 / direct API | Intelligent but not notable; extremely slow and far too expensive. | 205,710 | 3,215 | $0.72 |
| ChatGPT Astra Latest | Medium speed, token-efficient, highly intelligent; cost similar to Fable. | 278,603 | 1,536 | $0.83 |
| Claude Fable 5.1 | OK speed, very expensive, medium token efficiency; solved in ~53s. | 397,806* | 2,857 | $0.85 |
| Gemini Flash | Refused the reverse-engineering task. | N/A | N/A | Not reported |
| Thinking Machines Inkling Small | Got stuck immediately and was terminated. | N/A | N/A | Not reported |
+----------------------------------+--------------------------------------------------------------------------------+-----------------+-----------------+-----------------+
```
The latest Deepseek-Flash model was the most impressive of the open models. Intelligent, fast, cheap. What more do you want? I think the Xiaomi MiMo - Flash model did well, but it might be a fluke, given how poorly the Xiaomi MiMo Pro model performed (it wandered off on a million side quests).
What's also interesting is how competitive Luna is to the lower end Chinese models. It's quite a bit more intelligent, burns a lot less tokens, is plenty fast and very cheap. After doing this experiment, I don't think it's a foregone conclusion that the American AI labs will be having the bottom of the market eaten away by the Chinese models. It seems they (at least OpenAI) can compete on the bottom end as well.
Finally, the benchmark's results reinforce a few things people have been saying about Sonnet and Terra, namely, they're not competitive for a range of tasks. For easier tasks, Haiku/Luna are plenty capable. For harder tasks, Opus and Sol are so much more intelligent, that they can get them done quicker and cheaper. Maybe there is a place for them in the "Opus designs and Sonnet implements" workflows, but in that case, why not "Opus designs and Deepseek implements?"
PS Code harness used was Claude Code. $20 in openrouter credits was used to access the models. Kimi K3 and Deepseek were also accessed directly via the API and then compared to Openrouter. Kimi K3 specifically was much faster on Openrouter, and the difference in output tokens to get the same results for Deepseek/Kimi makes me think the Openrouter providers are hosting significantly quantized versions of the model. Goes to show that Moonshot.ai's acceptance testing of providers is a useful thing, otherwise your model might look artificially bad when it's hosted as heavily quantized by providers who dont care about quality, just top-line cost and speed.
For the past few days I've been trying to use rea to reverse engineer my Samsung Galaxy Buds 4 Pro to get seamless functionality working outside of the Samsung ecosystem with no luck (the buds 4 pro require Samsung devices + a Samsung account for seamless to work, I only have a Galaxy phone + my macbook). I think we're still far away from having something that works that well.
Dog bark and baby crying detection are advertised features for SmartThings integrations. Door bell and door knocking are also included, this is surfaced directly if you hook the TV or sound bar into SmartThings.
I had to check this and it's true. Not saying Samsung or 3rd parties don't take advantage of any data they can collect, but the story the tweets tell is super biased against them. Here, the detection seems limited to specific pattens required for a particular feature.
If OpenAI really banned him for that reason, that would be pretty stupid. If you can no longer use an LLM for reverse engineering without worrying about your account, that drastically reduces its value.
In my efforts to reverse engineer the LG firmware to validate Gamers Nexus claims, I got a warning very quickly.
I appealed that quickly and successfully.
Then I applied for daybreak blue and haven’t had any issues since.
It also a matter of phrasing. Say what you want to achieve not how if it is approaching defensive or cyber stuff.
Also state you are working on your own devices and software. (If that is the case of course.)
Yes a personal subscription. Individuals can also request it. only blue for defensive questions. Red for offensive is gated to organisations that are proven to be using it for protecting.
"Why you bother about local AI? LOL with the $ you spend for <insert hardware here> you can buy 5 years of 20x account". Yeah. 5 years where you are told what you can do.
> if your samsung TV model doesnt have a microphone, it will look for connected devices with one, such as a soundbar or a connected bluetooth device. this gets shipped off to bixby.
The AI in the TV was listening for dogs barking and babys crying, a gross invasion of privacy. Possibly this could have been for ad targeting purposes.
With projects like REA, it's going to get harder and harder for companies to get away with things like this.
I'd love to see the major social media apps reverse engineered and analyzed in this way. I can only imagine the dodgy practices that'll be found if this happens.
To what extend do these "smart" products state how, when and what data they are collecting and selling?
Isn't it high time for legislation that at least makes it clear on the box what kind of sensors and integrations it has and how bad the product is privacy wise?
Also, shouldn't corporate spying disguised as automated content recognition and classification be limited or outright be banned?
And all this connected garbage should by law be able to function with the cord cut and without registration, and if an update negatively changes the advertised product characteristics the customer must be allowed to return it for full paid price even if it is 10 years old. Force the manufactures to make statements on the "box" about the features, the support duration and level, and the privacy characteristics.
Why don't politicians care about this?
They don't care because of several reasons. 1. They get paid not to 2. They are told not to.
Alot of this spying ends up producing valuable data for different agencies to purchase which otherwise they cannot legally obtain without a warrant. Sure some of it has some commercial value, but a lot of it is probably more valuable for domestic spy operations.
I have heard enough send 18 billion to Israel
I miss the days when these low effort comments were frowned upon here.
People complaining about orange site degradation are probably as old as the site itself.
Remember guys, if they are reading it for wrongthink, they are reading it for industrial espionage.
Anthropic and openAI just announced they’re spinning out competitors to their biggest customers… right after ingesting all of their IP and contracts. Do the math.
(Or don’t, because they’ve been caught stealing math research)
> because they’ve been caught stealing math research
No. They haven't been caught stealing. Tristan Buckmaster publicly expressed surprise that the OpenAI model has used the niche approach he pursued (with AI assistance). 10,000 agents exploring various approaches make it much less surprising.
The "coincidence" defense never works for random individuals so I won't buy it when it comes companies worth billions which will lie, cheat, and deceive to make even more billions. They've been caught lying and stealing again and again. After so many years of this I think any trust put in them is either blissful ignorance or willful misdirection.
Interesting. Could you share a link to this announce?
Announcement
This looks very bad on OpenAI.
What is next? Power and water company suspending your account if your home and car are not streaming enough telemetry.
I am glad that open models are closing on frontier.
Claude is absurdly sensitive to anything that touches cyber, and threatens bans if it gets a whiff of a pentest going on. it wouldnt surprise me if openai had a similar approach.
Yesterday I joked that claude gave me a cyber stop for asking how to make the anime girl video for an exploit, but it was a thing that happened. Meanwhile im trying to convince people around me that having grok nudify people isnt "fine" just because grok says it is. Its wild where the lines a drawn.
I'm in the new CVP, and even Mythos gave me a cyber stop just for creating an MCP server that can remotely connect to my Steam Frame to run commands. It must've thought I was trying to create a centralized malware or something.
Not even sure why Mythos even still exists when it's identical to Fable in every way, including refusing to do stuff that looks like it could ever be repurposed into unapproved use cases.
Tastelessness aside, this seems perfectly rational to me. Unlike reverse engineering for the purpose of developing security exploits, "nudification" doesn't meaningfully expand the capabilities of malicious actors with access to free online porn and Photoshop.
Actually violating one person is fine, but doing anything associated with a specific industry that has applications for harm and help is not?
That's like saying it's fine to beat someone up and take their lunch, but it's not fine to learn how locks work.
It's fine to imagine what your lunch looks like, even if that makes you feel that your dietary privacy has been violated.
That's one of the weirdest takes on here in a while.
I can develop security exploits myself just fine without the use of AI. Many people don't have the capability to use Photoshop and free online porn to deepfake someone else they might or might not personally know into a realistic video...
With Kimi K3.1 Pro, today WAS the day they actually caught up.
I’ve been saying this for the past few months - once the inflection point happens (such as today), it’s only a matter of 2 years when American closed models won’t be able to catch up with Chinese open models
Iirc They specifically trained them to not be useful in cyber stuff. But the point still holds
… IF chinese models stay open.
Bad on OpenAI? The account got suspended.
This reflects horribly bad on Samsung!
The headline is misleading, but the "article" is only five screenshots of Twitter posts. Someone used AI to decrypt the Samsung TV's firmware and found always on listening, classification, lots of privacy nightmares. Then their (the hacker's, not Samsung's) OpenAI got suspended for "Cyber Abuse".
Right, but still: OpenAI provided a service and it's up to them to determine acceptable use. Samsung sells a $3000 (or actually $2999,95) TV and you get big brother watching you.
This is again will be very funny race to watch. Right now you can fairly do lots of reverse enginnering or "hacking" with off the shelf Chinese models, and they are very capable (GLM3 for example) which can be run locally if you stomach the hardware. Tools like rea are also becoming more popular. If openai continues to be pedantic, they will lose the technical base given how close is the gap getting.
The company I work for has a plan B to switch to alternative, self-hosted models if providers pull off strange moves, such as drastic pricing changes. We have already switched from one provider to another once, and we do not want to get burned again. We do not reverse engineer, just regular development.
Until Kevin quits and you need to modify his work
Have you tried Qwen3.8-Flash-Next? I found it significantly better in anything I tested than GLM5.3-Flash, but the thing is you either use Alibaba cloud or you host it yourself (no lesser quant than unsloth 4.5 or mx4/nvfp4 and kv no more quantized than int8).
It is great if you have enough RAM. I have enough to run it, but not enough for prefix cache. 27b is so much faster with long term tasks.
The Qwen3.8-Flash-Next model is ok for reverse engineering. It can definitely solve simple tasks, but it gets easily confused, and burns tokens like crazy.
I've been sick with stomach flu the past few days, so I put together a small benchmark for all the top AI models where I ask them to reverse engineer a very simple small crackme (https://www.crackmes.one/crackme/6ab4ce5195b976f8f1300b54) using Ghidrasql (https://github.com/0xeb/ghidrasql)
(As an aside: I think representing tools for AI as virtual SQL tables is much better than traditional MCP servers, and that this approach deserves a lot more attention. Props to Elias Bachaalany for thinking out of the box and putting it together).
The results of the benchmark are below. It's sorted by cost, but I added notes about each AI model for myself. The notes try to talk about intelligence, speed, and cost. Intelligence is a subjective measure which I assign to each model by reading its thinking tokens and watching its tool calls. You can definitely tell which models are more intelligent, even on this small benchmarks, by their tool use, false starts, and how easily they fall for the simple misdirection in the crackme.
(gist is more readable on mobile) https://gist.github.com/markoglasgow/8812c19b6fbffed30b2661b...
```
```The latest Deepseek-Flash model was the most impressive of the open models. Intelligent, fast, cheap. What more do you want? I think the Xiaomi MiMo - Flash model did well, but it might be a fluke, given how poorly the Xiaomi MiMo Pro model performed (it wandered off on a million side quests).
What's also interesting is how competitive Luna is to the lower end Chinese models. It's quite a bit more intelligent, burns a lot less tokens, is plenty fast and very cheap. After doing this experiment, I don't think it's a foregone conclusion that the American AI labs will be having the bottom of the market eaten away by the Chinese models. It seems they (at least OpenAI) can compete on the bottom end as well.
Finally, the benchmark's results reinforce a few things people have been saying about Sonnet and Terra, namely, they're not competitive for a range of tasks. For easier tasks, Haiku/Luna are plenty capable. For harder tasks, Opus and Sol are so much more intelligent, that they can get them done quicker and cheaper. Maybe there is a place for them in the "Opus designs and Sonnet implements" workflows, but in that case, why not "Opus designs and Deepseek implements?"
PS Code harness used was Claude Code. $20 in openrouter credits was used to access the models. Kimi K3 and Deepseek were also accessed directly via the API and then compared to Openrouter. Kimi K3 specifically was much faster on Openrouter, and the difference in output tokens to get the same results for Deepseek/Kimi makes me think the Openrouter providers are hosting significantly quantized versions of the model. Goes to show that Moonshot.ai's acceptance testing of providers is a useful thing, otherwise your model might look artificially bad when it's hosted as heavily quantized by providers who dont care about quality, just top-line cost and speed.
For the past few days I've been trying to use rea to reverse engineer my Samsung Galaxy Buds 4 Pro to get seamless functionality working outside of the Samsung ecosystem with no luck (the buds 4 pro require Samsung devices + a Samsung account for seamless to work, I only have a Galaxy phone + my macbook). I think we're still far away from having something that works that well.
Dog bark and baby crying detection are advertised features for SmartThings integrations. Door bell and door knocking are also included, this is surfaced directly if you hook the TV or sound bar into SmartThings.
I had to check this and it's true. Not saying Samsung or 3rd parties don't take advantage of any data they can collect, but the story the tweets tell is super biased against them. Here, the detection seems limited to specific pattens required for a particular feature.
If OpenAI really banned him for that reason, that would be pretty stupid. If you can no longer use an LLM for reverse engineering without worrying about your account, that drastically reduces its value.
In my efforts to reverse engineer the LG firmware to validate Gamers Nexus claims, I got a warning very quickly. I appealed that quickly and successfully.
Then I applied for daybreak blue and haven’t had any issues since.
It also a matter of phrasing. Say what you want to achieve not how if it is approaching defensive or cyber stuff.
Also state you are working on your own devices and software. (If that is the case of course.)
Was this through a personal subscription? Last I looked into daybreak access, if required an organization based account.
Yes a personal subscription. Individuals can also request it. only blue for defensive questions. Red for offensive is gated to organisations that are proven to be using it for protecting.
https://chatgpt.com/cyber
> If you can no longer use an LLM for reverse engineering without worrying about your account, that drastically reduces its value
...for a niche of customers, from whom the revenue is probably entirely negligible to OpenAI.
Not saying it's right, ethical or anything like that, but it's not "stupid"
It creates a lot of uncertainty. Especially if you depend more and more on them in your day to day life.
If your always-on agent hears to wrong thing, you could lose access to things you depend on. The same reason makes GCP very unattractive.
"Why you bother about local AI? LOL with the $ you spend for <insert hardware here> you can buy 5 years of 20x account". Yeah. 5 years where you are told what you can do.
5 years where you can get arrested if the AI tattles to the feds.
Claude Code for coding and DeepSeek + opencode for reverse engineering at this point.
DeepSeek will break the binary into code and Claude will search for problems. Now you have most of exploitation chain.
from his twitter account:
> if your samsung TV model doesnt have a microphone, it will look for connected devices with one, such as a soundbar or a connected bluetooth device. this gets shipped off to bixby.
Is this not the expected function when you use bixby?
(Not saying people should use bixby mind you)
Source: https://news.ycombinator.com/item?id=50193579
This is why you buy a display panel and not a smart TV.
I'm curious if someone has started cataloging these cases. These spying TVs seem like a common industry practice.
I believe GamersNexus were talking about doing more exhaustive testing across different brands after their video on LG models specifically.
A video on vulnerabilities in Asus routers was released yesterday: https://youtube.com/watch?v=kYnwCx2SNYY
But if you are talking TVs specifically, I expect they will get back to that too.
https://mrbruh.com/asusrouter/
Nice.Just don't buy any tech home appliance that was made after 2022.
The AI in the TV was listening for dogs barking and babys crying, a gross invasion of privacy. Possibly this could have been for ad targeting purposes.
With projects like REA, it's going to get harder and harder for companies to get away with things like this.
I'd love to see the major social media apps reverse engineered and analyzed in this way. I can only imagine the dodgy practices that'll be found if this happens.
so is the hardware mic-off button on Samsung TV any good?
If we read the tweet - the TV then connects to whatever device it can like a home theater, other speakers and uses its microphone if it has one
No libre agents and LLM's?
The tile references on-board ai. “On-board ai” here is automatic content recognition. Imagine my disappointment
AI is simply an if/else statement in disguise, you didn’t know that?
way to miss the point.
so glad i bought a dumb tv.
i remember going store to store in search of one.
gg me, that was a smart move.
What a load of shit. Samsung TV always listening if you enable listening, why would they do this? /s
How dare they run a classifier locally on the TV, they should upload it uncut to Czechia like Amazon did with their ring cameras!
the point is the setting does nothing, now go back to your claude session