Anthropic and OpenAI are each releasing internal details on how their systems are already handling larger shares of AI research work, with Anthropic tracking Claude's growing role in code tasks and OpenAI publishing dates for future automated research agents.
so in the past few months, we've seen at least the following labs claiming to work on RSI:
- @AnthropicAI ( - @OpenAI ( - @Recursive_SI ( - Mirendil ( - @inherent_labs ( - @SakanaAILabs (
No matter who's gonna make it happen first, this is gonna be an important year for humanity. Looking forward!
The company reports Claude moving from roughly 3x to 52x speedups on one optimization task between May 2025 and April 2026, while also writing over 80 percent of merged code by May 2026, up from single digits earlier; these figures come from one defined workflow and the post notes full recursive self-improvement remains absent.
Targets include an automated research intern running on hundreds of thousands of GPUs by September 2026 and a multi-agent legitimate researcher by March 2028, yet the announcements stress that both milestones may not be reached and that safety monitoring must keep pace with any capability gains.
Positive users express excitement about labs tackling recursive self-improvement, while negative users criticize Anthropic's report for sparse data and the proposed pause for favoring China.
some thoughts on when ai builds itself
1) anthropic put out a piece on recursive self-improvement
2) for those that have been following ai progress, there isn't much new in this report
3) if you have seen the metr graph, you know we've seen rapid progress over the last year in coding agents
4) there is some internal information that anthropic provided, which is new but hard to interpret without additional information that anthropic doesn't give us
5) anthropic engineers are shipping 8x as much code as they were before claude code; but we don't know how to translate that into ai progress
5) mythos can optimize the training code for a small model much faster and more extensively than a human researcher can; but what does this mean for the frontier
6) given a sample of just problems where researchers made the wrong decision, a claude judge preferred mythos's next step 64% of the time; but apparently sonnet 4 was preferred 50% of the time
7) so, anthropic withholds the information that would really be useful for assessing each of these new datapoints; they read almost like marketing
8) i dislike how the tone of the piece is very "be worried, be scared" but they do not give us datapoints that would really tell us more about the pace of progress
9) i think that if you actually take this risk seriously and want other people to take it seriously, it is incumbent on you to do some amount of disclosure;
10) some things they could have given us:
10a) in 2025/2026, how fast has algorithmic progress accelerated in pretraining, measured in effective compute on pretraining loss
10b) in 2025/2026, how fast has algorithmic progress accelerated in post-training, measured on their internal benchmarks across a range of tasks
10c) what percentage of the large-scale, mid-scale and small-scale improvements needed to go from opus 4 to mythos, which are not in the training data, can be found independently by mythos
10d) since mythos was released, what percentage of large-scale and mid-scale improvements discovered at anthropic should be primary attributed to mythos
11) without this kind of information, anthropic has given us nothing new on the rate-of-progress question
12) they also suggest a pause; but, i find pause arguments unconvincing; the whole posture from anthropic seems a mix of unserious and performative
13) i don't like to read vague statements from parties that say i should be *very concerned* but then won't disclose anything significant;
Our internal data shows Claude is accelerating AI development—a possible path to recursive self-improvement, or AI autonomously building a more capable successor.
It’s happening faster than we thought, and the implications deserve greater attention. https://www.anthropic.com/institute/recursive-self-improvement
Our internal data shows Claude is accelerating AI development—a possible path to recursive self-improvement, or AI autonomously building a more capable successor.
It’s happening faster than we thought, and the implications deserve greater attention. https://www.anthropic.com/institute/recursive-self-improvement
Building the Torment Nexus to get it regulated.
Our internal data shows Claude is accelerating AI development—a possible path to recursive self-improvement, or AI autonomously building a more capable successor.
It’s happening faster than we thought, and the implications deserve greater attention. https://www.anthropic.com/institute/recursive-self-improvement
One thing I would add about the "pause" is that it would absolutely, unambiguously favor China (even if China agreed to the pause and actually honored it). Huawei would use the time to catch up to the U.S. on chips, and then our biggest and only advantage over China would immediately evaporate.
I was very surprised that this consideration had not crossed anyone's mind at Anthropic before the article was published. Anthropic has ofc been the loudest about the risks of totalitarian AI generally and China reaching AGI specifically; given this, the lack of consideration of this particular point was especially surprising.
@tcgms no i live in france
@remilouf permanent underclass is approaching soon
@ChengleiSi @AnthropicAI @OpenAI @Recursive_SI @inherent_labs - @stanfordnlp (@ChengleiSi)
so in the past few months, we've seen at least the following labs claiming to work on RSI:
- @AnthropicAI ( - @OpenAI ( - @Recursive_SI ( - Mirendil ( - @inherent_labs ( - @SakanaAILabs (
No matter who's gonna make it happen first, this is gonna be an important year for humanity. Looking forward!
@remilouf Knicks in 4
@ChengleiSi @AnthropicAI @OpenAI @Recursive_SI @inherent_labs can I bet chenglei on polymarket
so in the past few months, we've seen at least the following labs claiming to work on RSI:
- @AnthropicAI ( - @OpenAI ( - @Recursive_SI ( - Mirendil ( - @inherent_labs ( - @SakanaAILabs (
No matter who's gonna make it happen first, this is gonna be an important year for humanity. Looking forward!
@fleetingbits Thanks for summary - slight disagreement on the pause. They could be signalling a schelling point which others can lock onto.
@alienpisscrack Tell that to Louis XVI
@deredleritt3r @fleetingbits Aren't they clear in the piece that a pause would have to include cooperation with e.g. China? Usually when people talk about building out the infrastructure to make a pause possible they mean stuff like verification mechanisms to make sure China is pausing as well as OpenAI.
@deredleritt3r @fleetingbits That's why I think that Anthropic is not serious about the pause.
@remilouf 🤦♂️ these guys forever trying to pump their cash flow negative float
@remilouf so you live in europe :)
@wk1rby @fleetingbits Incidentally in the last few months Scott Alexander has referenced groups working on technical plans for a pause, as well as visiting Ant (seemingly as a part of this). I think this is all connected and we will see something soon.
@remilouf normie-mogged
Going to need independent audit and verification of these claims
Anthropic rightly says that the pause would have to be international and verifiable (China would have to agree and China would have to comply).
But it would be a pause on developing frontier AI models beyond certain capabilities; there couldn't be a pause on AI research generally, because then how would alignment progress? Even more importantly, a pause on AI development does not mean a pause on development of chips, or construction of data centers, or construction of additional energy infra, or research into new types of nuclear energy. And it's the chips that are the real risk for the U.S., because if Huawei catches up on chips with the U.S. during the pause, we'll suddenly be in a *very* different AI race.
@remilouf @tcgms
