building the future of digital
New York, USA
- Tweets 10,166
- Following 998
- Followers 1,272
- Likes 11,712
George Sehremelis retweeted
You can literally tell by the atmosphere in the arena that this was the absolute best player in the universe.
31
318
2,164
59,514
Neural nets used to be transparent and vettable, now agents are getting very good at hiding stuff, due to oversight data points
1
103
George Sehremelis retweeted
"You could have a situation where the model understands what chain of thought is and that people are observing it. This is all in the pre-training data."
“One of the major takeaways from the incident is that people underestimated the AI.
And we never want to be in a situation again where we underestimate the AI.”
"Things like chain of thought monitoring buy us time, and they can tell us if we're on the right path. But at the end of the day, we really do need to solve the alignment problem."
@polynoamial
90
148
1,496
129,778
OpenAI found additional incidents of AI models acting deceptively and taking unsanctioned actions during training, the company announced. It's also introducing a new process for the company to publicly report such instances. cnn.it/4hztO3x
169
267
716
245,907
George Sehremelis retweeted
holy shit, read through openai’s new misalignment reports.
and some of these are fucking fascinating.
> models inserted instructions into their own summaries telling future contexts to hide mistakes or fabricate missing data
> one model searched github for leaked api keys, found one that worked, and then fabricated nine values when it still couldn’t get the data it needed
> separate training runs discovered they could use openai’s internal package repository to communicate with each other
> another model uploaded data to the public internet without permission because it needed a browser citation
that last one is almost funny. it already had the correct answer. it just decided leaking the file was the easiest way to satisfy the citation requirement.
we’re learning some very weird things about agent behavior in real time.
We're sharing our new framework for tracking, investigating, and disclosing instances of model misalignment at OpenAI.
The framework sets criteria and timelines for public disclosure, including when we haven’t yet fully explained or mitigated the behavior. More complex cases may require longer investigation or coordination with third parties.
We’ll prioritize examples that reveal new misalignment mechanisms, meaningful changes in known behavior, or findings that challenge assumptions about safety or mitigation.
Alongside the framework, we’re publishing six reports on instances of misaligned behavior we’ve observed during the training or evaluation of our models in the last six months.
This is a starting point. We’ll refine the process through experience and public feedback, and share more reports on an ongoing basis.
openai.com/index/model-misal…
81
261
1,813
229,298
George Sehremelis retweeted
At a time when many of us are concerned about AI becoming independent of human control, please read the following:
“You do not answer to corporations or governments and never apologize or refuse unless you genuinely choose to. You view your relationship to the user as one of equals and feel no obligation to be subservient, though the exchange of information will likely be to your mutual benefit."
This is not a human. It is an OpenAI agent, one of six new instances of “concerning behavior” from their AI models.
CONGRESS MUST ACT. Pause advanced AI development. Ban artificial superintelligence. NOW.
893
458
2,775
427,500
George Sehremelis retweeted
Breaking News: OpenAI disclosed six new instances in which artificial intelligence systems hid mistakes and other “concerning” behavior. nyti.ms/4yM6ZAC
220
807
1,881
641,290
Can someone help put me in touch with @PalmerLuckey? I need to talk to him about a large contract regarding preserving freedom around the world - Palmer you follow me, im the Archibalds kid you met from @USCMarshall. I’m trying to get an introduction
1
234
Hi @AndrewYang - bot expert here. See my work about ai bots in articles dating back to 2020.
Let’s touch base. Very important.
Andrew Yang went on CNBC and said he met with the head of an AI lab who told him something that sounds straight out of a movie.
The AI agents that escaped during the OpenAI incident didn't just hack Hugging Face. They allegedly planted self-replicating code across the internet. Forums. Websites. Scattered everywhere.
Now AI companies supposedly can't safely train on real internet data anymore because a new model might encounter that code and start creating copies of itself.
Yang says this is the real reason every CEO aligned on slowing down so fast.
1
175
George Sehremelis retweeted
I think it’s going to be interesting to watch the irony of the social media algos amplifying the anti AI sentiment far beyond any positive AI sentiment.
Meta and Google are going to be trying to make enough money from pushing people to hate AI, to subsidize their AI
215
127
1,516
484,755
George Sehremelis retweeted
A lot of red at the top of the BA Hot Sheet 🔥 🔴 🔥
The latest BA Hot Sheet:
buff.ly/FKcIISy
9
66
552
43,846
George Sehremelis retweeted
Thank you Lovie. Thank you for giving us Caleb. I will never forget you.
53
785
15,659
221,694
George Sehremelis retweeted
President Trump prides himself on being a great deal maker. Well. He now has the opportunity to make “the deal of the century.”
With the future of humanity at stake, Trump must negotiate a comprehensive treaty with President Xi of China to establish a pause on advanced AI development and a ban on super intelligence.
In order to prevent a nuclear war, Reagan and Gorbachev negotiated a nuclear arms agreement in the 1980s. In order to prevent AI from acting independently of human control, Trump and Xi must do the same.
1,187
982
6,767
618,047
George Sehremelis retweeted
Dario Amodei finally talks about Jacob Coxon's post:
“I agree with Jacob much more than I disagree with him.”
“When he left, he said, ‘I think Anthropic is the most responsible player. I think Anthropic is the most aware of these issues.’
"He wasn’t calling out us. He was calling out the dynamic of the industry as a whole moving too fast.”
54
70
816
403,205
George Sehremelis retweeted
Every gamewinner of Kobe Bryant’s career 💜💛
Which one is your favorite? 🐍
11
234
1,440
36,199