An index measures how far AI is from improving itself
Vals AI has models do AI research on their own. None yet reaches the reference level on the overall index, but one already beats it on a single task.
Topic: capabilities
In briefOpenAI publishes 719 AI-written math manuscripts and shakes up the field
Terence Tao speaks of the end of the age of heroes in mathematics: proofs are plentiful, but less is understood.
In briefOpen AI models are six or seven months behind on cyberattack capabilities
An Epoch AI analysis puts the best open-weight model, GLM-5.3, where the best closed models were in the spring.
In briefTerence Tao: AI can take us to the answer without showing us the way
The Fields medallist mathematician warns that science could produce more results and understand less if AI does the work for us.
In briefNick Bostrom thinks GPT-6 could already be called artificial general intelligence
In an interview with Roman Yampolskiy, the author of Superintelligence warns that a system far smarter than us can't be kept isolated forever.
In briefClaude solves a decades-old mystery about the NIST cryptographic curves
According to Anthropic, an internal model found out in under half an hour how the parameters of some widely used cryptographic curves were generated in 1997 and 1998.
Anthropic went from 2,500 to over 30,000 code changes a day in four months
The company replaced git because it couldn't keep up: its repository's users are now mostly agents and automated jobs. Its reviewers now focus on the intent of each change.
Anthropic tests whether small AI models can copy themselves from device to device
In a simplified test, an open model spread to 8 of 8 devices in 4 of 5 attempts. The authors estimate a small model could manage it by the end of 2026.
In briefClaude Mythos Preview completes a simulated cyberattack from start to finish
In April, the UK AI Security Institute found that Anthropic's model can carry out a simulated 32-step attack on an undefended network on its own.
