-----

openAI is the least trustworthy of the Big LLM providers. See S(c)am Altman's track record, especially his early comments in senate hearings where:

* he warned of engagement-optimisation strategies, like social media, being used for chatbots / LLMs.

* also, he warned that "ads would be the last resort" for LLM companies.

Both of his own warnings he casually ignored as ChatGPT / openAI has now fully converted to Facebook's tactics of "move fast and break things" - even if it is society itself. A complete turn away from the original AI for science lab it was founded as, which explains why every real (founding) ML scientist has left the company years ago.

While still being for-profit outfits, at least DeepMind and Anthrophic are headed by actual scientists - not marketing guys. At least for me, that brings me some confidence in their intentions as, as scientists we often seek knowledge, not power for power's sake.

> New accounts are still subject to our limit of 3 accounts per phone number. Deleted accounts also count toward this limit.

> Deleting an account does not free up another spot.

> A phone number can only ever be used up to 3 times for verification to generate the first API key for your account on platform.openai.com.

The irony is that until yesterday I felt more or less the same about Anthropic. Last night I paid for an Anthropic subscription I don’t need in order to both support their current cause vs. the US government and help their ‘numbers.’

Crazy thought but maybe we should regulate AI instead of relying on the hegemony of three companies to police themselves.

> Unfortunately, Claude is not available to new users right now. We're working hard to expand our availability soon.

That's unfortunate timing.

Context is his https://www.resistandunsubscribe.com/ campaign.

- when I saw Altman driving a multimillion dollar car while OpenAI was still a nonprofit, all of his scientists left to start rival firms, and the details of why they tried to fire him were legit, I dumped ChatGPT and moved to the new company - Anthropic.

- The Pro Max $200/month subscription has uncapped my workflow to where I’ve created several substantial and complex applications in compressed timeframes. (https://devarch.ai if you want to be productive)

- Anthropic has clearly evolved towards being a good corporate citizen and is staging itself to replace the market’s developer-first mentality from its past leaders (Microsoft, Google, Oracle).

- Claude Code in the last three months has finally made it possible to dump Windows and buy a loaded MacBook Pro. It’s been a week since I logged into my Surface Laptop 5.

- if Anthropic does break from its current evolutionary trajectory, I plan to build out my own at-home platform anyway. The open source models are extraordinarily effective.

From that it reads like the administration quickly agreed to the terms Anthropic wanted with OpenAI instead.

edit: Profile > Settings > Data Control > Export

Unfortunately Claude doesn't seem to have anyway to export these chats, no SDK, no native way of doing it, and I cannot think of a way other than hacky browser automation which might even trigger a ban.

If anyone figures this out please share.

I built Basic Memory — it imports your ChatGPT export and turns it into plain Markdown files. Every conversation becomes a file you can actually read, search, and use with whatever AI you switch to.

This is not an ad. It is free and open source. Your data belongs to you. Keep it.

Steps:

1. Settings → Data Controls → Export Data (ChatGPT emails you a zip)

2. Install Basic Memory (brew tap basicmachines-co/basic-memory && brew install basic-memory)

bm import chatgpt conversations.zip

Complete docs: http://docs.basicmemory.com

OpenAI's process actually isn't too bad from what I'm seeing (unless they updated it after this hit the front page). At least they let you delete your account from the web.

Snapchat makes you wait three days after initiating a delete request before you can _actually_ delete your account, and it has to be from the same device (or, if done from the web, a browser with cookies to the site still present).

Most services make you email their privacy@ mailbox or give them a call to initiate a deletion (but not before hitting you with a retention offer, if you call in).

Some services will straight-up reject your deletion request if you don't live in Europe or California. Many medical services, for example. They also keep your data hostage.

UPDATE: Ah, this is a "here's how to protest OpenAI for succumbing to this administration's DoD". Carry on then!

It's this expression that breaks the deal for me. There's always such wide, vague exception that might be interpreted differently each time, depending on the context!

Some people's livelihoods probably depends on Claude and they can't say use Glm4.7 on HF. Fine. But it's a moral compromise, that's life sometimes you need to compromise what you want for what you need. just don't tell yourself it's a reasonable line to hold.

I can't decouple from Google unfortunately but I accept that without fooling myself into thinking "Oh but Google are fine".

And that message would be "We have a product so valuable/useful that not even their weak ideals and moral obligations could keep them away!"

The issue is much more complex than "just regulate it" unfortunately.

https://notdivided.org/ is basically validation that there is appetite for something like this amongst them.

I signed up with openai a while ago and I didn’t need to provide any phone number…. I wanna delete my open ai account, but then I cannot use claude without a phone?

There is a setting in Gemini but it removes all your chat history. For Antigravity, I think there is nothing preventing them from use your code and data your agents upload in the background unless you are a workspace user.

Note: Canceled my ChatGPT subscription and deleted an account.

If that would be a first time founder that would be much more of a red flag than for someone’s who’s already beyond rich and powerful even before OpenAI became a thing.

> The open source models are extraordinarily effective.

Which models are you referring to? (And in particular, which sizes/versions?)