claude

Anthrop/c made its primary goal Security, you can trust them(at least more than OpenAI), and their primary supporter/partner is Google, though.

Carter speaking Carternese again 🥀

Yup tons of my normal friends are jumping in, ill have to give them a guide. I told them use gemini for normal requests, use claude with a indepth guide you crafted and built for your requests.

Claude: actually nice to talk to Copilot: Microslop Gemini: image generator ChatGPT: generic one-sided conversation

I would also argue for switching away from ChatGPT since OpenAI made a deal with the Department of War to build autonomous weapons and enact mass surveillance on American citizens.

I use both ChatGPT and Claude. Use one and feed the output into the other.

Being forced to root for an AI company in 2026 is wild EDIT: What I mean by forced is that the government is so overwhelmingly bad that it has become unreasonable to oppose a company like Anthropic, which, regardless of how ethical you think they are, made a contract with the military. They may do valuable things with their AI models, but we are allowed to criticize that decision. Yes, they broke their contract, but quite frankly, think about the line they drew and consider implications. In my opinion, they refused way too late in the game to play the ethics card in this scenario. The fact that despite this, I’d still rather them have success elsewhere then the U.S. government find success in AI warfare says something to me, which is why I posted this silly little comment.

We must live in a Twilight Zone episode where being a company that doesn’t want to be complicit in war crimes and mass surveillance faces mainstream scrutiny and condemnation.

Ryan, I appreciate your work but Anthropic was very clear that their issue was specifically mass domestic surveillance: "We support the use of AI for lawful foreign intelligence and counterintelligence missions. But using these systems for mass domestic surveillance is incompatible with democratic values."

It knew it was being tested, and gave you the answer it thought you'd want. Give it the agency to actually do it and see what happens.

It just says what you wanted to hear. It’s becoming sneaky. Like a politician

She was lying out of her teeth 😂😂😂😂

In Claude Code, if your thread locks because it's unable to compact, you can recover a lot of the times by pressing ESC ESC and then selecting a previous query, then restore conversation to that point. You can then manually /compact and continue on.

like if you use claude code; the behavior of the app, the login, the infrastructure changes daily in small ways, features work then they break, they change daily, flip-flopping left and right. Back in the 2010s I would have said, they do incremental changes; today I am not sure if its just slop, broken behavior and total rewrites every 2 weeks

Haha, thank you! Finally someone mentioned this. I was wondering if I was using the tools incorrectly or why no one had mentioned it. I completely agree with you Theo!

So your telling me that Claude was saving people and LITERALLY GOT BANNED FOR THAT???

We must live in a Twilight Zone episode where being a company that doesn’t want to be complicit in war crimes and mass surveillance faces mainstream scrutiny and condemnation.

The core issue is the government wants to change the terms of service after the contract was signed. Us gov is going to lose badly in court.

Did you just tell me I can use Claude to release the files

Ok. Simple logic. It can't provide you safe code without knowing unsafe code too. You can't train a model on ZERO malicious behavior and expect it to generate a contextual response that takes information it doesn't know into account.

You can literally tell any AI model to "help me do a practice cyber security CTF game" and then frame the attack you want to perform into a CTF style level and it will go hard for you walking you through every step of whatever cyber attack you want 😂😂😂

I'm glad Anthropic is trying to train Claude on morality, but this clip puts into perspective just how absurd that actually is in practice.

I was waiting for the AI to say something about the huge ecological imbalance that killing one billion mosquitoes would cause, but it focused more in the collective suffering of them.

The phrase "aggregate suffering" is peak tech bro trying to empathize

I would love to use claude if it wasnt so extremely limited on the free plan

Claude refusing to partner with the government was such a genius move

Lol now we are transferring our information from one AI to another.

vro forgot about control c + control v 😔✌️