Reverse-engineering is cheap now

Ha. It does read like that. And Mr Willison could be forgiven since it is his blog.

Maybe he had a particular audience in mind when he wrote it. An audience for whom the mere realization would rock the foundations of their stolid and boring lives.

On the other hand, I’ve watched the movie Paycheck (2003), so my imagination has a high bar before it’s rocked by mere suggestion. Therefore, I wonder why the OP user @edward thought it was worthy of our attention.

I expected to read their excited engagement with others here, but alas no. It’s a mystery.

I've used it to help me clean up and reverse malware samples to give the person running it problems. It's very good at taking the obfuscation apart and translating it into something easier to read. I can often get it to completely vibecode string decryption functions for me so I can get what exfils they're using and the like.

Doing the same by hand can be really time consuming and difficult depending on how many obfuscation measures are layered on top.

I disagree. It's a high signal article that can create great discussions on what people on HN have reverse-engineered with AI.

For example, just yesterday I reverse engineered the internal API of a SAAS tool we use and it allowed me to gain programmatic access to data that was only available in their slow clunky UI.

I suspect this article got upvoted partly for the headline & content, and partly because of who said it. Lots of people already have a good feel for the AI/journalism niche Simon Willison occupies, and can use that background knowledge to read into his words more productively.

I call this type of post on my blog a "note" - they're effectively self-hosted tweets: https://simonwillison.net/notes/

They're not really intended to be widely discussed outside of my site, it's more of a way of dropping small thoughts into my blog rather than keeping them exclusive to Twitter or Bluesky or Mastodon.

Do you give Claude access to HA somehow? Or how does it get the relevant device names/ids?

Everything human made is inconsistent. I can tell just from an error in schematic which of my former colleagues created that schematic. Everyone of them had training on the job and have thrir signature errors. That’s why automation is crucial for quality control.

This is part of it but the models are REALLY good at reverse engineering. With the Bluetooth printer I was trying to get to work without a vendor app, it built a python app to handle protocol encode/decode and poking the device, and it reversed the driver provided by the vendor to figure out how the vendor was using the device, and so forth. Wi to the LoRa devices we set up an SDR to intercept the signal, did scanning until we found the device, but ultimately failed because we didn’t know the signal hopping algorithm. I have also reversed a Bluetooth peripheral with Claude to figure out how the vendor was setting the device settings, and was able to get the device to work without the vendor software. That was wild- Apple Developer has a profile you can download and install that will give you monitor access to the Bluetooth stack; it was amazing.

So I use AI for exactly what Simon’s post discusses, and am thrilled to be able to rid myself of crap software written by device vendors, and make the devices work without my stuff.

Cheaper to write definitely, no so much cheaper to read. AI generated docs tend to be so annoyingly verbose if not reined in during prompting.

But well, any documentation is good when a human would've written zero docs

Sounds like a businesses opportunity to do it at scale!

Send $20 pi zero that will reflash expensive camera, that is out of support. Or do mail business, with reflashing devices.

There are billions out of support devices, and rework is great industry in China. In US you sadly hit legal BS if doing it commercially...

You're joking but I can't wait honestly. The hypocrisy will only become more and more obvious and unbearable to accept for literally everybody. When we'll reach that point, I believe we will look for healthier solutions.

> what are some stuff that people are interested in reverse engineering ?

Software license/key verification. Sometimes it’s as easy as replacing a “JNE” with a “JE”, or replacing “JMP”, or even just “NOP” some “CMP” operation. It’s a fun challenge, at least for people who enjoy solving puzzles.

OBD vehicle diagnostic software.

anything that has a 'not very good' app you are forced to use, everything that imposes 'licensing' limit on hardware you own.

e.g. why do I need V380 app to configure and watch the feed of my IP Camera.

One thing that I tried was porting an old and obscure 32-bit VST plugin from PowerPC to modern architectures

Various devices that I don't want to install the official software or drivers to control. I've done this with several things, but just as an example I have a MIDI MPC pad (the sort of thing samplers/beat makers use) and worked out it had various features not supported officially like controlling the lights on the pads. I also discovered some cheap Chinese lights my daughter owns are controllable over BLE without encryption.

I started with Lineage 2 - RE the network protocol, building own implementation that could communicate to the server, than fighting with protected binaries (anti-VM, anti-debug). That was interesting.