Trying to make things that vaguely resemble games (and the tech to make them work). Recently released corpo/ghost as @InterleavedSys
Joined August 2011
- Tweets 7,291
- Following 86
- Followers 151
- Likes 19,202
598 Photos and videos
Exclusive: I have seen OpenAI's audited financials for 2024 and 2025.
In 2025, OpenAI had $13.07 billion in revenue and $34 billion in costs. $867 million of its revenue came from SoftBank, and $303 million came from Microsoft.
wheresyoured.at/exclusive-op…
148
633
3,542
1,594,351
Arvīds Kokins retweeted
Jun 15
New: Researchers quantify how easy AI search is to manipulate. 13 words buried in a random Reddit comment can poison AI search results. They suggest this is not easy to stop: "The way you can attack these systems is so much dumber than you think it is"
404media.co/it-is-trivially-…
1
89
225
9,321
Arvīds Kokins retweeted
Jun 12
Those who celebrate @elonmusk's $1 trillion fortune need to be reminded of a simple and vital truth:
That there is a fundamental tension between extreme wealth and the very possibility of democracy.
551
1,058
3,085
940,819
Arvīds Kokins retweeted
Jun 12
Tried out the UK government’s official new AI chatbot for jobseekers.
Its advice: work in the US.
Quick, which public service can we add AI to next
153
667
8,941
405,360
Arvīds Kokins retweeted
Jun 12
🔴 Please sign this petition for the AI Copyright Transparency Act forcing companies to disclose their training data!
They've been clearing hurdle after hurdle thanks in large part to the petition, let's help them clear the last few!
7 Nov 2025
🔴CALL TO ACTION to fight AI! Takes 10 seconds!🔴
Please sign and share this petition to send to legislators in California to support the AI Transparency Act. It would force AI companies to provide a list of work they trained on to IP holders. ANYONE can sign!
Link below!⬇️
2
71
230
5,878
Jun 12
if you waste any money on these fucking consultants, you're being scammed
KPMG report on AI found riddled with AI hallucinations bit.ly/4uAnE7w
38
Arvīds Kokins retweeted
Jun 10
Generative AI is just a giant billboard that says “I didn’t care about making the thing, but I still expect you to engage with it. I bet you can’t even tell, you idiot. Fuck you.”
I’ve never felt more deeply committed to making handmade things than I do now. ✏️
22
920
6,087
57,050
Jun 10
holy shit, this is a sharp turn
how it started (only a month ago): arstechnica.com/ai/2026/05/a…
Jun 9
More AI-generated code doesn't make your team faster. It might actually slow you down.
52
Arvīds Kokins retweeted
Jun 9
Comedy of AI errors: Lawyers on both sides of a case misuse AI, cite fake hallucinated cases.
At the show-cause hearing, "each of the attorneys expressed embarrassment and apologized to the Court."
"Neither of them verified the legal authority output by AI before filing their briefs."
In other words, there were two clients who basically were paying for ChatGPT (or whatever LLM) to argue against itself.
Neither of them realized this until the court issued the OSC!
Fines all around (between $1,000 and $3,500), other sanctions include pro hac status revoked, disqualified from practicing in the district for 2 years, among other things:
60
309
1,509
264,284
May 31
for an entertainment product, copilot isn't especially entertaining tbh
May 30
“Ex-Microsoft exec says the company blew it with Al, as it did with mobile”
"Not even 3% of paying Copilot users use it even when it's pre-deployed right in their faces”
The Microsoft 3% problem. See Word and Excel features.
1
77
May 28
union busting at the wikimedia foundation
consider sending them an angry email (especially if you're a donor)
medium.com/@jakeorlowitz/wik…
43
May 23
checked a different video on that same channel and sometimes there are no issues with the pronunciation of "WWE" but sometimes the same thing can be heard, I wonder what causes it
An automated A.I. WWE news channel on YouTube tries to pronounce "WWE" as a word and ends up going insane!🤣
2
1
88
Arvīds Kokins retweeted
May 21
Hello @JeffBezos, since you question the results of our studies on the unfairness of the US tax system, please allow me to remind you of the main conclusions of our work, the most comprehensive research to date on this issue.
May 20
Yes, the United States has the most progressive tax system in the world. The top 1% pay 40% of taxes, the bottom 50% pay 3% of taxes. We can make it even more progressive by zeroing out taxes on the bottom half. It’s a small amount of the total tax revenue but very meaningful to people in this group.
173
1,852
9,974
2,162,699
Arvīds Kokins retweeted
May 21
Berkeley law has introduced a new, much stricter AI policy
law.berkeley.edu/wp-content/…
139
816
4,714
963,851
May 18
the one I got is funny too
google even adds that highlight for the thing you're meant to focus on, except it's complete nonsense as well
1
66
May 18
also "remove definition" interpreted as a command reminds me of "emit continent" from MGS2
a search engine LLM has no way of removing anything, not to mention erasing an idea from the entire world
36
May 15
"AI" slop in research is finally getting a pushback
going after incontrovertible evidence is 1) something everyone could do in 2023; 2) indicating the situation is bad
though I'm curious how they'll handle incorrectly claimed "co-authors" to avoid this policy being weaponized
May 14
Attention @arxiv authors: Our Code of Conduct states that by signing your name as an author of a paper, each author takes full responsibility for all its contents, irrespective of how the contents were generated. 1/
49
Arvīds Kokins retweeted
May 13
💬 Digitālā neatkarība ir arī valsts drošības jautājums.
Iniciatīva aicina valsts pārvaldē un skolās kā prioritāti izmantot atvērtā pirmkoda risinājumus, lai stiprinātu datu drošību un efektīvāk izmantotu budžeta līdzekļus.
👉 manabalss.lv/i/4115
1
2
87