Post

Conversation

Jake P. Taylor-King
@wildtypehuman
This is legitimately terrible for OpenAI --- but also Anthropic --- and it relates to their work in drug discovery. I read this as: prompts will be monitored and potentially acted on if we think there’s something scientifically and/or commercially valuable being disclosed. Now, the big problem for Anthropic is that they are increasingly seen to be competing with their customers. Outside of SF, many people also think of OpenAI/ChatGPT and Anthropic/Claude as somewhat interchangeable companies and products --- aka, if OpenAI do something, surely Anthropic will do it too? As so much IP relates to drug targets and people disclose these to Claude, I think we will see a lot of pharma companies stop using these tools or put up a lot of internal guardrails.
Quote
Thomas Wolf
@Thom_Wolf
wtf is this way to handle mathematicians work and scientific communication TLDR: Leven and Tristan worked over several months on one of the Millenium Prize Problems with various AIs to reach final interesting results. OpenAI apparently heard about it in the last days and x.com/deaneyang/stat…
Image
84.2K
Views
View quotes
David Watson 🥑
Post your reply

sneedster
@flzjvididns
this has been their open M.O. for a while and in the future it will probably extend beyond that. their goal is to clone and steal anything made by human hands. soon it will encompass everything from books to games to research papers
87
AGTP
@AGTPinsights
The privacy concerns regarding proprietary research are massive. We actually went deeper on this here:
Quote
AGTP
@AGTPinsights
Mathematicians Tristan Buckmaster and Levent Alpöge just triggered a fight over who really cracked these fluid equations. Here's what you need to know. Buckmaster, an NYU math professor, and Levent Alpöge spent about a month working with AI models, mainly Claude and Codex, on x.com/deaneyang/stat…
Image
1.4K
Chris
@chrisw96722402
The people who invented a plagiarism machine are secretly competing with their customers? I’m shocked, shocked I tell you!
92
Richard She
@Richard_She
I know some academics that are paranoid about this. On the one hand, they promise us Free Claude, and supposedly no data leakage. On the other, why would you be making lists of sophisticated users pushing the boundaries of science... Early FB and SNAP certainly had looser
Quote
Aaron Ring
@aaronmring
Replying to @RippaSatss
The terms of service for teams are clear that they don't train on team or enterprise data by default. And that users own the inputs and outputs... I read the fine print 🧐
2K
Ergo
@FebreoToHerJuly
You will not stop asking God for answers even if your God is cruel. IP and Copyright are on their way out. Human flourishing doesn’t care about the old ways of capturing surplus value
502
Kamil Pabis
@Aging_Scientist
Jake, let's not jump the gun. There is so far zero evidence that prompts were monitored. In fact, this was immediately denied. It is alleged that OpenAI had tipped off by his mail as to which approach is most promising. This is the strongest claim of malfeasance in this whole
598
stottpie
@stottpie
Anecdote: Claude helped me develop a patent. Even went through a checklist, and helped me with proof that it was not an ai invented technology (which would not be patentable) and created documentation showing the time stamped artifacts that were used by the ai and aided, not
112
Matt Creigh
@MattCreigh
This is why it's so important that Nations recognise this for what it is; an arms race where independent means of model creation ought to be the goal
383
Jac Nom
@NomJac25845
Then the models are not so impressive. Because the solution was already there and they just stole it while during publication/finishing phase.
574
AllTheTokens ∞
@AllTheTokens
I'm disappointed this has come up for OpenAI. It's obvious that Anthropic do this and have been doing this forever, because they are liars. But OpenAI seemed like they were trying not to be evil like Anthropic.
557
cesar leon
@ivan_leon78
Anthropic? what they did? The problem here is OpenAI they were stoling the ideas, what was the problem with Anthropic?.
30
alexandriemartinez
@alexandriem7
If you’d never once used the phrase “Navier–Stokes” before today and you’re suddenly tweeting about it like you’ve spent your life doing computational fluid dynamics, you’re a fucking cunt and you produce nothing of value.
36
_cjk
@_colinjk
The onus is on the labs to prove systematic controls make peeking impossible; or if peeking is required, for legal or other reasons, that it's logged, and the logging cannot be tampered. Trust requires third-party auditing and full transparency.
392
Aeon
@aeon208
Ya'll, Sam Altman literally had someone killed. Elites that eat children are pushing AI. Why are we surprised by the ethics of these "companies"?
70
Lalit Dhalia
@lalit_dhalia
The amount of money they would be paying to these companies, they might as well just set up their own open source models, on-prem, engineer salary included and still have money left over AND keep everything private. As always, the problem is management.
166
#M U G A I N B I O
@smchatter1
AWS and cloud providers spent years proving their internal security and encryption to guarantee security. Models are app level processes that private client work is absorbed by the model host is obscene.
50
Danny Moore
@dannymoore_ni
Another reminder that the public models are industrial scale IP harvesting machines... their primary function. Same for many of the intermediaries,.. one of the headline reasons for the SpaceX acquisition of Cursor was their database of coding prompts and responses for Grok
391
Seth BRKV
@SethBRKV
So they basically just go, hey, Claude/Codex sift thru all the sessions you had with customers, find most unique and promising ideas and let’s appropriate them!
7