How do NLAs work? An NLA consists of two models. One converts activations into text. The other tries to reconstruct activations from this text. We train the models together to make this reconstruction accurate. This incentivizes the text to capture what’s in the activation.
RESEARCH
-
Natural Language Autoencoders Explain AI Activations
By
–
Natural language autoencoders (NLAs) convert opaque AI activations into legible text explanations. These explanations aren’t perfect, but they’re often useful. For example: NLAs show that, when asked to complete a couplet, Claude plans possible rhymes in advance:
-
Anthropic Research on Natural Language Autoencoders
By
–
New Anthropic research: Natural Language Autoencoders.
— Anthropic (@AnthropicAI) 7 mai 2026
Models like Claude talk in words but think in numbers. The numbers—called activations—encode Claude’s thoughts, but not in a language we can read.
Here, we train Claude to translate its activations into human-readable text. pic.twitter.com/pMLsxM2VAONew Anthropic research: Natural Language Autoencoders. Models like Claude talk in words but think in numbers. The numbers—called activations—encode Claude’s thoughts, but not in a language we can read. Here, we train Claude to translate its activations into human-readable text.
-
Multi-agent research adds verifiers for code execution and repair
By
–
This builds on our existing research on multi-agent systems, with a key addition: verifiers. Planners spawn workers that write code and verifiers that run it. If verification fails, the planner spawns a new worker to fix it.
-
Coordination defects cause majority of multi-agent LLM failures
By
–
Pay attention to this one if you build multi-agent systems. Coordination is as important as prompts or agent architecture. Multi-agent LLM systems fail in production at rates between 41% and 87%. The majority of those failures are coordination defects, not base-model
-
xAI and Anthropic have opposite compute problems despite large GPU fleet
By
–
The xAI / Anthropic compute story is not about one company having GPUs and the other wanting them.
It's that they have opposite problems. xAI reportedly runs one of the largest GPU fleets in the world. Yet according to The Information, its recent model FLOPs utilization was -
Scale AI publishes a new benchmark for refactoring agents
By
–
Scale AI published SWE Atlas Refactoring Leaderboard, a new benchmark that evaluates agent capabilities of restructuring the code. > It requires agents to produce twice as much lines of code than SWE Bench Pro. > Claude Code with Opus 4.7 tops the leaderboard followed by
-
Research on many-shot prompting limits
By
–
More examples don’t always mean better results. Our team's research explores the real limits of many-shot prompting and test-time adaptation across models and tasks. Read the paper
-
Hugging Face Hub crosses 4,000 public RL environments, largest platform?
By
–
The @huggingface hub just crossed 4,000 public RL environments! Does it make us the largest platform for RL envs or are there bigger ones? Let us know what we can do to improve on the topic, it's early and we're excited to keep growing and support more! https://
huggingface.co/spaces?categor
y=agent-environment&sort=trending
… -
AI’s Extensive Impact on Surgical Workflows
By
–
The impact of AI for surgery (pre-op, intraop, post-op) is going to be extensive
A new review https://
frontiersin.org/journals/scien
ce/articles/10.3389/fsci.2026.1783803/full
…
Our previous review @NatureMedicine https://
nature.com/articles/s4159
1-024-02970-3
…