giancarlostoro

Agree, it makes sense, a lot of people are catching on to LLMs for reverse engineering finally, but its been possible. LLMs can and will eventually flawlessly dissect models. I experiment with Claude, I ask it tasks I know are possible, but have no idea the solution, it can reverse engineer obfuscated projects faster than most talented people I know in that field (talking about games now not AI). I heard a developer say their latest release would take at least a month to figure out the obfuscation and encryption, Opus 5 did it in like... 15 minutes of talking to me (I did throw it the right questions / suggestions). I do that all in the name of seeing just exactly what is possible.

Claude has been able to do reverse engineering for nearly a year if not longer, as well as full source code rewrites, if you know how to do a proper rewrite, luckily for me, most of my career has been rewrites.

> how you could prune harmful capabilities from models and build a more efficient training system

In theory, if OpenAI or Anthropic wanted to release a new open weight model, they could scrutinize it to remove the most horrendous things from the model.

They are notoriously difficult to contact if your company is banned, or someone within your company is banned. That's it, poof, gone, everything you have.

That is a weird stat, I would assume it "fell" because all the insane amounts of free compute that was going around died down around those years, and some companies finally made AI policies. My employer (in one of those fancy Fortune lists) still is not using it in a meaningful way for development.

They're basically trying to make themselves into the likes of Azure, GCP and AWS. Now you might think that there's no way Oracle will ever be that big, but half of their revenue for the 2026 fiscal year is from their cloud infrastructure, which, trust me, surprised even me.

Oracle is the last company I'd setup shop with.

Because they're just taking a tweet by some antisemitism troll account (that is apparently "satire") at face value.

Apple not updating their maps since October 7th for that region is a documented thing. So much so, when Google updated theirs (who also froze theirs) they compared Apple Maps vs Google to see what's changed. The tweet is just rage bait.

https://english.elpais.com/international/2026-09-29/gaza-bef...

You can 100% delete those without needing to update map data, my statement still stands, there's plenty of mainstream news sources that confirm that Apple has not updated their mapping data for the region since at least 2023. Google updated theirs in 2026 and one news org immediately compared Google Maps in the region vs Apple Maps in the region:

https://english.elpais.com/international/2026-09-29/gaza-bef...

> Why would Apple make a remark about this? You can literally go look for yourself and see they've done this. As another commenter pointed out, they almost immediately removed the names of ethnically cleansed towns from Southern Lebanon as Israel wiped them off the map.

They have not updated the satelite images since 2023, this is a known documented thing, news sites have literally compared Apple Maps to Google Maps when they finally updated their images for the region in 2026. Apple can remove roads if it knows they're no longer around, it doesn't mean they have to update satellite images, which has been the case.

> Says who?

Literally browse posts by that account, it doesn't take very long to see antisemitism throughout the account. ...Why are you touchy about that?

Every Windows laptop I have ever had where I installed Linux on it, I swear its lifespan is infinitely longer than if I kept Windows on it. My wife has Macs that are older than my oldest laptops / desktops I can still turn on. Windows is just abysmal for hardware.

Lmao a colleague who works elsewhere said some goon at work created an AI leaderboard. He spun up research agents. Top of board achieved.

The issue is all the "replacements" are very studio specific. If they really were smart, they'd pull together resources, launch a new company under the competing studios support, and give themselves better pricing for games sold. They'd get more into their pockets, and a small cut to keep the new platform sustainable.

Are we just making assumptions? I've built up a few different mapping / GIS related projects over my career, and the satellite imagery can change on a whim, sometimes its spot on, other times its dated, the zoom can also affect it as well, you zoom in and its from years back, you zoom out slightly, and its from last week.

Has Apple made any remark about this to confirm? It would not surprise me if Apple just isn't even aware of this changing on their app. (For those unaware, they source their map data from various sources / vendors, as does Microsoft's Bing Maps.)

> I've also not seen anyone else make this claim.

That's what I call Steve Jobs style PR / Marketing you deflect the problem by making another problem to make it seem like you're fully aware. "You're holding the phone wrong"

> but if it'd cheaper the authors don't go into

Merging a model and a LoRa is insanely cheap, so assuming it didn't use any GPUs to do inspection etc, and just looked at a file on a FS, then I would be willing to say... It's cheaper than training a LoRa if that's the case, assuming Astra didn't take several weeks?

> No human can go edit a brain to become better at math.

For now...

At some point you have to rip out the bandaid, if they did everything possible before operating on actual test subjects, then I don't personally have an issue.

At which point from testing with non-animals do you go from simulated surgeries with robots to actual surgeries though? I don't want it being done unethically, but how do you promote the effort / research ethically?

If we can even have robots that can perform critical surgeries with total precision and save more lives would that be bad?

Note this was from March so its not implying something happened today. Still useful information.

You know what happens to technology when enough people stop using it?

Dont worry, if OpenAI and Anthropic have their way and everyone loses their jobs to AI, we're all becoming luddites.

Setup an unprotected network that is isolated and you fully control, see if it connects at all.

So your TV just connects to random wifi networks? That sounds a bit insane.

That sounds like a TV manufacturer is going to be having one hell of a lawsuit if they're breaking into other peoples wifi networks.

I just dont connect mine to the internet, hook up an Apple TV to it.

Im always afraid of Japanese toilets, will I sit on it as the robot uprising starts?

It didnt, just small enough, which is good, its way more useful to train small purposeful models than to use an entire frontier LLM.

I have been using it to train a few BERT models.

Those are rookie numbers, you need to pump those numbers up…

Thats when you build a giant bus and select what you need out of it.