What is interesting to me is that stripping the C2PA data is easy, but faking it is hard.
You can resave the file and the "made with Claude" signal disappears, but you cannot make a random file pass as Claude-made without Anthropic's signing key. So the useful guarantee is one-way. No signature means almost nothing.
The goal of C2PA is that cameras will start to emit C2PA credentials. You will then have 3 situations:
* C2PA confirms a photo is authentic
* C2PA confirms a photo is AI generated
* C2PA missing, you don't know.
I reckon we will only see "C2PA missing" being treated as suspect in select situations (perhaps Reuters will require C2PA from their photojournalists, for example)
Camera C2PA can never meaningfully confirm that a photo is authentic, it bears about as much credence as EXIF metadata. It's like saying the existence of DRM confirms that a movie hasn't been pirated.
C2PA cryptographically guarantees that the bytes came from a hardware/software signer and that the signed payload has not been modified since that signature was applied.
So no, C2PA is not as easy to spoof as EXIF.
And no, the existence of DRM doesn't validate the integrity or the provenance of the bytes.
This is not the case with current systems, but could future systems be designed to be tamper-evident in a way that makes it impractical to sign fake images or extract the signing key without leaving evidence on the device?
If that were the case, I can imagine a subscription service in which you get a camera for some specified period of time, and then return it to the company that sold it for them to verify the camera hasn't been tampered with. Then the company could publish a list of which keys (unique per camera) have been verified to not be tampered with. Maybe this wouldn't stop everyone, but now the person trying to fake images has to re-do the process every so often and I imagine it's more expensive to avoid leaving evidence.
This might be too impractical to work, and it would be bad for privacy, but maybe for some people the tradeoffs actually would be worth it, someday. For now, I assume there are much cheaper and easier ways to detect faked images, at least for expert humans.
I assume the signing key is different from each camera unit(not only model), so if a picture of you winning lottery in US capture by a camera sold to someone in Thailand, it would be extreme unlikely to be real.
"Faking" it is trivial. You don't need their signing keys when you can just ask them to sign whatever you like. Upload your own file with the prompt "present this file back to me again, as-is".
One place where I find the Flash models surprisingly bad is Google Search's "AI Mode".
A recent example - I searched for how to unsubscribe from Pearson emails. Google Search "AI Mode" confidently gave me a sequence of steps along the lines of Settings > Profile > Email preferences > Unsubscribe.
Of course, I looked for an unsubscribe link before asking Google. None of those options existed. The correct answer was there is no way to unsubscribe through the account, so I just blockthe emails instead.
I've run into this pattern quite a few times. AI Mode seems to make up things all the time.
I think that's just a limitation on the size of the model. I'm pretty sure that they use a pretty small model in those summaries to save money, which naturally makes them a little less smart.
>We will not send marketing emails to a user who has opted out of receiving them. Any marketing communications we send will include an unsubscribe link at the end of the email.
I don't think this is AI's fault. This is Pearson's publishing incorrect information and the only way to really know they are a bunch of lying assholes is to have an account and try to unsubscribe from it.
It's obvious that the Google Search AI Mode encourages the model to give an answer without spending unnecessary cycles investigating deeply.
They also heavily encourage keeping the context short. For example, it will remove the option to start a new turn after a small number of turns, depending on the topic.
It definitely makes things up all the time, but it gets it right surprisingly often. I really like it.
It only makes sense if we are measuring raw capability per dollar. Once you factor in noise, physical size, idle power, battery etc, I think the calculation changes.
For a home sever that's mostly idle, it is almost like commuting in a pickup truck.
My server is an extremely quiet Xeon workstation: unlike 1U servers, workstation are typically quiet. The good thing with a Xeon workstation is that it supports ECC RAM only and there are plenty used. For there's no market for used Xeon systems, so you can find stuff like 22-cores / 44-threads CPUs for $25 or something.
Some mod Xeon systems to use regular PC PSU like Be Quiet! ones and Noctua fans, etc. but my HP Z440 workstation is so quiet already I didn't need to.
In my case it's an headless server and is sits between the vacuum cleaner and the cat's food: there are literally only two cables connected to it, the power cable and the ethernet cable.
I can live with the footprint of one tower.
I run many VMs and containers inside those VMs: Git server, pastebin, NAS, containerized/VM-isolated AI agents, etc. but I don't need it to be up 24/7. I basically only need those when I'm working. So electricity usage ain't a big concern to me as it's more like running 8/7.
I don't feel like I'm commuting in a pickup truck: I feel like I have the perfect machine for the task.
It averaged 69w. The Mac Mini I replaced it with uses just under 5w.
A Mini is possible the worst bang for buck. Also, the horror that that is Desktop Docker can’t be over emphasised. I know there are better ways of using Docker on a Mac, but having got something working I can’t face rebuilding it.
My home router is a Lenovo ThinkCentre Tiny with a Core i3-8100T with a dual 10Gbit SFP in it (1x DAC, 1x 10GBase-T) and it idles at 14W, routing 10gbit full bore 24W. I got it for $80
I also have some Fujitsu office machine I picked up for $50 as my "I need to run some random task on a physical linux box" machine with an i5-7500 and it idles at 15W
It’s sluggish, it’s incredibly bloated with all the services that Docker (the company) wants to sell you and the UI is badly designed.
OrbStack is my go-to when it comes to running docker on Mac. It also uses macOS native containerization APIs so containers will be a bit more performant than they would (at least use to) be on Docker Desktop.
Another good approach if you’re fine without a GUI or want to bring your own is just ‚brew install docker‘.
The main gotcha I encountered with orbstack was with pulling down internally built amd64 images and trying to run them on apple silicon. I just remember constantly running into issues and going back to docker desktop as a result, even if their solution is janky in its own way.
Trying to get it to open on startup and start containers in a particular order is a little finicky. But the real hell is trying to get it to start containers after the network shares have connected.
I’ve got it working, but compared to doing it in compose, it’s awful.
I think the problem is concentration. AI now absorbs almost all of the oxygen from every category. A story about new chips becomes an AI story. Cloud pricing becomes an AI story. Even personal projects get pulled into the same orbit.
If I am archiving PBs of data for 10+ years, I don't want to rely on a WASM interpreter being available and performant in the future just to read a file. I want a dead-simple, heavily documented byte specification like Parquet.
Additionally, putting the decoding logic inside an WASM binary introduces an active execution layer into what should be a cold storage.
WinRAR format does include RAR VM bytecode as part of the archive to achieve state of the art compression in media files. it was sandboxed and well accepted by everyone.
the same sandboxing capability exists for WASM as well.
it is actually better for long-term archival: you dont need to carry decompression program, since it will be a part of the archive file itself
It's a selection bias issue. The categories you have listed are essentially web services wrapped in an app shell. Of course they need the internet. Consider these examples:
- Photo/Video editors - Snapsheed, Lightroom, Video trimmers etc.
Most of what you've listed I do on my computer not my phone. And some of your items do require internet for useful purposes (at the very least backup and syncing for notes/passwords).
We (software engineers) get better outcomes from the same algorithms by improving data flow, constraints, instrumentation etc. (Better) prompting, retrieval, context engineering etc seem like the LLM equivalents.
The model weights haven't changed but the system is making more use of the capabilities already present in the model.
If your implication is that humans are deterministic, then that's laughable. We like to think highly of our mental processes, but humans work in different ways than AI and it shows. They are good at different things.
You can resave the file and the "made with Claude" signal disappears, but you cannot make a random file pass as Claude-made without Anthropic's signing key. So the useful guarantee is one-way. No signature means almost nothing.
reply