DeepMind has found a way to hide a signature inside an AI-designed protein without breaking it
The target is gene synthesis screening: a provider that can detect the mark can tell where an ordered sequence came from.
AI Research — Page 2
Coverage of artificial intelligence research: notable papers, new architectures, interpretability work and results that change what models can do.
The paper's authors include OpenAI's chief scientist, an Anthropic co-founder, Microsoft's chief scientific officer and the Turing laureates Geoffrey Hinton, Yoshua Bengio and Andrew Barto.
Stanford and Caltech drove a humanoid through unfamiliar rooms with no trained action layer at all
The robot explores first, the run is rebuilt as a digital twin in Isaac Sim, and the model picks a skill and a target on every step.
Claude computed a nine-loop scattering amplitude that nobody in the field had reached
Matt von Hippel asked AI companies to take a frontier amplitudes calculation on an academic's budget. Two Anthropic physicists gave Claude one prompt and told it to keep going while they slept.
Stanford and Nvidia have opened a decision-only model that answers by matching, not by writing
The weights are on Hugging Face under Apache 2.0, and the trained part is two small projection heads on a frozen Qwen3-8B encoder.
All stories
- Two AI models broke Enigma messages that had defeated cryptanalysts since 2005
- Forecasters gave under 25 per cent to the AI results that actually happened, an accuracy review finds
- The price of a fixed level of AI performance has fallen about 47 per cent every quarter since 2023
- Anthropic says Claude found an unknown enzyme system in virus DNA, and nobody yet knows what it does
- Nine mathematicians, including Edward Witten, will advise OpenAI on how to release results it says its model already has
- A paper removes the central orchestrator from a multi-agent system and runs it with 1,024 workers
- Give an agent your inbox and it starts recommending pricier flights, a study of 325,000 runs finds
- A paper argues agents that rewrite themselves are overfitting, and that holding them back makes them better
- Anthropic says Claude rewrote more than 30 open biology models and made them about four times faster
- A paper shows a model provider can pad its own output tenfold, and that you can catch it from outside
- Alibaba open-sources an abdominal CT model that outscored 23 of 26 radiologists in a reader study
- Google DeepMind launches an essay platform on AGI, and says its pieces are not Google's view
- Salesforce says it more than doubled an agent's browser success rate without touching the model
- OpenAI says work done outside your job description is becoming a regular part of how people use ChatGPT
- In 34 of 37 countries surveyed, people expect AI to destroy more jobs than it creates
- Mathematician Andreas Thom says OpenAI misled him about his private ChatGPT chats
- Twenty-five Fields medallists say the race to solve problems is damaging mathematics
- Anthropic publishes a model of three AI economies, and lets you set the dials
- OpenAI publishes its Navier-Stokes proof and will not claim the $1m prize