[giancarlostoro]

> On the bright side, maybe someone can get Claude or some other LLM to figure out how to crack it; and perhaps even vibe-code an alternative app.

LLMs are great for this, though the more people use it for blackhat style things, the more I fear they will lockdown LLMs which are useful for reversing things that are legacy as heck and abandonware.

[sanex]

I listened to an interview with the guy who nearly cured his dog's cancer with ai and he expressed concern multiple times that going public with the story would cause the labs to cut off the ability.

[hypfer]

For reference for other readers, here's the primary source: https://x.com/paul_conyngham/status/2036940410363535823

Also, thanks for mentioning! I wasn't aware of this

[gaiagraphia]

Thanks for this. Absolutely incredible read.

I'm really hoping these new tools will empower a new generation of people. It's sad that there even needs to be a term 'citizen scientist'.

[SturgeonsLaw]

Wow, that is one of the best use cases for AI I've seen. I wonder what the cost of all that was.

[smallpipe]

What is it with the AI bros that seemingly can’t write two pages without their precious LLM doing it for them. It could be a genuinely cool story, but I’m just put off by the LLMism everywhere

[hypfer]

Tbf in this case it might also just be that the guy talked _extensively_ over years with those LLMs, shaping language.

At least nothing tipped me off for that specific text. What did you see?

[Reason077]

Nearly?

[Dylan16807]

It greatly shrank some of the tumors but not all.

[veeti]

I've been doing some light reverse engineering with Chinese models on Opencode and have found they are already far less reluctant. Just don't ask about Taiwan..