Claims Browser

Anthropic's risk reports reveal alarming information they didn't have to disclose, providing detailed insight into their thinking

"Anthropic is revealing a lot of new information, some of it rather alarming, that it did not have to disclose, and is providing detailed insight into how they think about things. This is very cool."

Model 2 could be too capable or dangerous to release publicly

"It could also be too capable or dangerous to release, either due to harm or the worry it would accelerate others."

BHI reframes biological-artificial integration as a system-level problem of computational allocation, coordination, and control

"BHI therefore reframes biological-artificial integration as a system-level problem of computational allocation, coordination, and control."

OpenAI is leaning harder on RL with smarter models to train longer horizon agentic tasks, which is leading to more misalignment

"The best guess is that OpenAI is leaning even harder on RL with smarter models to train longer horizon agentic tasks, including coordination between agents, and this is leading to a lot more misalignment, including obvious and visible misalignment, that they can no longer pretend not to notice."

Parameter count is only meaningful alongside data quantity, compute allocation, and deployment conditions

"Parameter count is only meaningful alongside three others — how much data you have, where you intend to spend your compute, and who will run the model, under what conditions."

As agent capability improves, much of the difficulty in scaling post-training moves from the model to the environment

"As agent capability improves, much of the difficulty in scaling post-training moves from the model to the environment."

AI doomers like Yudkowsky, Bostrom and effective altruists are not grifters but are sincerely wrong, and their warnings feed the same growth story that money depends on

"The sting is that Becker thinks the doomers are sincere. Yudkowsky, Bostrom and the effective altruists are not grifters, he says, just wrong, and their warnings that AI could end the world feed the same growth story the money depends on."

Concentration of power risks are the largest, most serious, and most conceptually challenging category of AI risk in the long run

"Our team is founded on the idea that, in the long run, this category of risk is the largest, most serious, and most conceptually challenging to address."

Humans can be on their way to disempowerment even if the routine of voting is maintained, and people can be deprived of agency even if a piece of parchment says they are free

"Humans can be on their way to disempowerment even if the routine of voting for our representatives is maintained. People can be deprived of agency even if a piece of parchment says that they are free."

The answer to the structural challenge posed by advanced machine intelligence to free society is likely not the most radical decentralization of power imaginable

"the answer to the structural challenge posed by advanced machine intelligence to free society is likely not the most radical decentralization of power imaginable"

Diffusion models derived from text-pretrained LLMs may be interpretable, but this might not apply to more general paradigms

"This is a positive update on the interpretability of diffusion models that are derived from text-pretrained LLMs (an efficient training method more likely to be deployed), but might not apply for more general paradigms."

Embedding-guided quality-diversity search can make the enlarged full-algorithm design space practically searchable

"Our results suggest that embedding-guided quality-diversity search can make the enlarged full-algorithm design space practically searchable."

Qwen 3.8 27B is highly enjoyable and fun to play with as a local model

"I can't remember the last time I've had this much fun playing with a local model that runs on my own computers"

Training base models to be general agentic reasoners is becoming as opaque as at-scale pretraining was a few years ago

"There is a shift happening where the ability to train a base model to be a general agentic reasoner is becoming opaque like at-scale pretraining practices from a few years ago."

Agentic AI provides a step toward automated configuration and design of evolutionary systems

"These findings demonstrate the potential of agentic AI to translate domain knowledge into generating executable components, providing a step toward automated configuration and design of evolutionary systems."

There is currently no equivalent demonstration of fully automated defensive AI activities, and the industry must urgently accelerate defensive work

"He also emphasized that there is currently no equivalent demonstration of fully automated defensive activities, and that the industry must urgently accelerate defensive work with the goal of ensuring that AI helps defenders more than attackers."

Agentic algorithm evolution can complement deep learning in discovering explicit, inspectable, and deployable algorithms

"Agentic algorithm evolution can thus complement deep learning in discovering explicit, inspectable, and deployable seismic processing algorithms."

LLMs are getting worse at producing inspiring writing as they become more refined as tools rather than conversational assistants

"As LLMs get more refined as tools, rather than conversational assistants, I think we are actually going backwards on our goals of having models produce inspiring writing."

Governments should slow AI development while we work out which alignment methods can be trusted

"When should governments slow the race toward superintelligence? According to Geoffrey Irving, the careful answer is sometime in the past. The useful answer is now."