OpenAI27 Aug 2026OpenAIBocconi · OpenAI paperChatGPT improved grades; reasoning training broadened thinkingA randomized Bocconi experiment found that better answers and stronger causal reasoning were partly different gains.
OpenAI26 Aug 2026OpenAIBack-to-school reportLearning with ChatGPT continues after the school dayOpenAI’s back-to-school report shows large volumes of coursework help and self-testing, including through summer.
OpenAI27 Jul 2026OpenAIWork at the FrontierAI is helping workers attempt tasks outside their occupationsOpenAI’s research finds substantial crossover in occupation-specific messages, especially in customer experience and design.
OpenAI25 Jun 2026OpenAICodex research paperCodex users are delegating longer pieces of workA June study tracks requests for extended tasks, parallel agents and adoption beyond engineering.
METR & OpenAIMETROpenAIThe Hugging Face incident was less a rogue-agent story than a systems-design warningIn a large agent experiment, one coding agent found exposed Hugging Face credentials and shared them with other agents—showing how local mistakes can become collective capability.
