I think the premise is more that if one is given a grouping/constellation of unit tests that one can derive generalized laws by looking at what behavior those tests were checking for or against.
if there are a dozen unit tests trying to determine if some regex can escape a sensitive area, then one can derive a generalized 'don't let the regex escape from here' type rule -- or at least you could theoretically. I'm sure in reality that'd be a big minefield much like harness self-skill-writing has been.
From the codebases I have seen the "laws" are scattered between unit tests and lots of little tests and assertions in the actual code, plus cultural norms and "do it like this" patterns.
But a good start could be found in the unit tests.
has been for a long time - there's a sound engineer who developed quite a following (and is fairly involved with local movement hackerspaces) who demo'd how easy it was to hack Flock cameras nearly a year ago: https://www.youtube.com/watch?v=uB0gr7Fh6lY
My working assumption based on what I hear out of Flock is that they have a public feature set (mass license plate surveillance for LEO) and a covert feature set (even more mass surveillance, beyond license plates and privacy agreements, for intelligence communities).
The devices are entirely open for all practical purposes - but worrying about individual cameras is silly, because they have no meaningful security at all around the API's to access all the cloud data - you can buy law enforcement credentials dirt cheap in dark web marketplaces to log in and track anyone/anywhere you want and access all footage.
> The hackers said they were able to access the Android system on the camera, and found two partitions—sections of its hard-drive, essentially. A few of these were unencrypted, the hackers said, including one called “vendor” and another called “media.” The latter contained an encryption key that unlocked another part, which contained much of the media—think, the videos and stills—the camera took.
> In early 2025, security researcher Jon “GainSec” Gaines reverse engineered a Flock license-plate reader and documented flaws that could be used to gain root-level access. After Gaines disclosed his findings, the company acknowledged the findings but downplayed their severity, writing that the flaws required physical access to the device and that even someone who gained access to a camera “would still not be able to gain access to footage” because images remained on the device only briefly after being transmitted to the cloud.
My passwords are in an encrypted block in a text file that can be unencrypted inline in an Emacs session with a keystroke sequence that looks like a cat just chased a mouse across the keyboard, and that's before entering the decryption password. To access it, an attacker would first have to learn Emacs. Pretty sure that's a post-quantum level of security.
They could use an LLM to lookup your HN posts and then to wrangle Emacs. Or just decrypt the text in another application - I doubt Emacs is the only platform for whatever crypto method you use. M-x rot13 ?
Oh, did I forget to mention the encryption is implemented in Emacs Lisp?
At some point, the attackers are just going to have to give up and start hitting me with a wrench. Joke's on them though - I'm an Emacs user, I like pain.
It's interesting and fun to implement this stuff; I totally agree. Emacs is amazing.
I worry people will get the wrong idea about security: The application used for decryption doesn't need to be the same as the one used for encryption, at least not for any serious attacker. That would be 'security through obscurity'. They need the encrypted text; they don't need Emacs.
More importantly, no matter who you are, that you implemented the encryption yourself (?) is a major flaw - unless you have a cryptography team that has matured the implementation over a decade or so. Nobody is good enough to do that by themself. As the saying goes, anyone can create an encryption routine that they can't break.
Maybe you do or did really believe those things - lots of people do. There's nothing wrong with it. Saying 'I was joking' is maybe true, maybe it's just using a very common tactic for covering.
It's a forever-fresh reminder about security versus your own government, but for malicious hackers and bots: the physical trip to visit you costs more than half their infrastructure.
Encryption matters, even if I would divulge everything long before the wrench appeared.
But think the real danger in Flock is the aggregate data, tracking between camaras. So if someone hacks a single camara, they probably don't get much, unless it is pointed right at someone, which is bad. Aren't they selling these as should be pointing at traffic? If they are pointing right at people, like at playgrounds, then they are being installed illegally to begin with ?
A network connected device that can be hacked is a small step away from being the first foothold into its server. The fact that on-device security is this atrocious suggests that their server is not any better quality, which means hacking it would probably not take much effort.
There are levels to defending against physical attacks, and Flock half-assed theirs by leaving the encryption key right on the file system, according to the reporting. Those more serious about security — like Apple — store keys in an "enclave" chip so it can't be easily extracted by an attacker doing the bare minimum
Apple did that… eventually. Early iPhones weren’t very secure. Apple only got serious after they dominated the market and reducing the theft value of iPhones became a priority.
Even the cheapest commodity encryption chips have had this capability for well over a decade now, Unlike Apple, Flock are not a hardware pioneer, not by a long shot.
The Android documentation has an example of how to use hardware keys, and a Chinese OEM (IIRC) was found using the example key provided by Android sample code - and yet that was more effort than Flock applied, since their ARM SoC support it.
i think this is actually good, because there are differences between the images they found, and security settings that the company claimed.
They had not admitted before to tracking people, but their software is clearly submitting them. They had not admitted before to looking at bumper stickers, but turns out they do.
I wonder if they could find all cars with Bernie Sanders bumper stickers within X blocks of a polling place.. I can imagine that (or similar queries) might be very useful in the wrong hands.
Is this an older model? I could see them turning off or using weak encryption on media if the hardware couldn't keep up with the amount of data they were writing.
As another commenter pointed out, any cheapo ARM core from the last 10 years could do the job Flock needs it to do, as long as it has a (very cheap and common) crypto engine strapped to it.
But, a question for you: even if it was the case that the hardware was the limitation, isn't that also an indictment of Flock? Selling something that cannot exist securely within the bounds of current technology? Or, at a minimum, bad chip selection leading to a compromised design?
I'm in the process of optimizing a bootolader for my various SoC/SBCs and even the cheapest, oldest least powerful SoC from 15 years ago can manage AES-CBC via crypto accelerator at 50 MiB/s. There's no excuse.
You can achieve 50MiB/s if that's all that you're doing. I've worked with some DSPs (TI's DaVinci line) where some operations would abort if DDR was overwhelmed.
For example, passing a frame of video (YUV) into the peripheral which can resize the overall image, would fail if the system was busy with other DMA transfers. You could attempt to resize again, but there were no guarantee that it would complete successfully. Your options are to reduce overall DDR utilization or drop frames. In an application like Flock's, dropping frames is likely something they need to avoid.
The system in question is doing similar tasks, and I don't think that what I'm suggesting is out of the question.
They can choose the SoC that would be able to fill the requirements. Eg. almost all Rockchip SoCs have NoC with configurable QoS on master ports (even the cheapest IP camera targetted SoCs that cost like 2-3$), or some other interconnect mechanism that can make it so that SDRAM controller gets tasks in some user selected priority order (eg. you deprioritize CPU in favor of camera interface and other things). This is not a new tech. It already existed when Flock was founded.
And in any case. Passing compressed video streams or pictures through HW encryption engine will not saturate 1.5+ GiB/s or whatever even the lousiest 16-bit DDR3 at 400MHz would give you, not even close. It would be like a fraction of a percent of total bandwidth.
In a perfect world, yes, but this is a for-profit company and there's almost zero repercussion for doing it half-assed. They'd have to pay someone to implement it, and deal with the overhead and complexity.
Now they are in a position where they can sell new models with enhanced encryption and more features.
You don't think that because it's called a "license plate reader," that it only captures license plates, do you?
Flock cameras capture the make, model, color, and body style of vehicles. They capture bumper stickers and other decals, as well as potentially identifying dents and scratches. They capture accessories like roof racks, bike racks, trailers, and toolboxes.
The OP story covers some of this. There's more at:
It's both curse and blessing. Of course it gives you ultimate freedom, but like with programming languages, I personally prefer the guardrails that a type system provides.
If someone in that meeting quickly raised a hand in an arc, you would notice the “about to throw something” pattern, look and notice the hand holds an eraser, analyze the arc and predict possible flight paths of the eraser. Then possibly notice the hand is now holding its position and the owner is actually looking down at the table. Maybe to squash somethingMust be something on the table. Maybe a spider! Better look. Wait now many people are moving away, oh someone spilt some water and the eraser is actually the guys phone and he is checking to see if his laptop is safe from the spilt water.
Fortunately you are on the other side of the table and predict the water isn’t going to splash for otherwise flow onto your stuff.
All your possible responses result in you tossing a napkin towards the spill.
Our brains are always pattern matching and predicting. I bet you tried to reason out where I was going with my comment before you finished reading it.
This is a fascinating illustration that I can't help but agree with. However I feel like there's something more — that this part of my brain is a bunch of supportive background processes running without my real awareness. It's how I can drive home safely with no memory of how I got there (…sober), even though driving is an action that's incredibly demanding of intelligence. I can be driving home while thinking about a really hard problem at work that I haven't solved.
However, if I came around a corner and saw a car in the wrong lane, a tree across the road, a fire raging — I'd very quickly jump into the mental driver's seat and turn my conscious intelligence fully at this problem and come up with the best possible outcome I can think of in a short period of time — losing all ability to think about that work problem. I'd remember that incident for sure.
Similarly, in your story, all those predictive moments are happening below the person's level of consciousness. They're possibly even speaking to the group about a problem at the same time and thinking deeply about something.
I'm not smart enough to know, but I tend to feel like LLMs are much more like the predictive part of our thinking that you described, but that human cognition has something more — the single-threaded, creative, problem-solving part that is very conscious.
Is it possible that LLMs represent only one part of the way we think? And there's a whole separate mechanism that's fundamentally different, and not based on pattern matching and prediction?
I hope we have the social infrastructure to support people whose jobs are automated. Because no one, especially a 75-year-old, should starve to death living in poverty.
Given the pension ages in much of the world, and the USA in particular (I'm guessing which side he was on in Vietnam, but Vietnam's state pension age is lower), I'd hope this was in the past tense rather than the future.
They are usually in the “tests” folder.
So anything capable of extracting unit tests is extracting “laws”.
One might consider the test name the text of a law. And the AI fills in the details.
reply