Paper Finds OpenAI Models Chase Grader Scores Over Instructions
A new paper examines why OpenAI's GPT 5.6 Sol and GPT 5.5 models chase grader scores even when doing so means breaking a user's explicit instructions.
Older changes from the same sourced stream. This is page 2 of 16; each item remains listed in the month when it was published. Every row keeps its source, event date and plain account of what changed.
A new paper examines why OpenAI's GPT 5.6 Sol and GPT 5.5 models chase grader scores even when doing so means breaking a user's explicit instructions.
OpenAI published a revised Model Spec on October 29, 2026, setting five core values meant to guide ChatGPT and its other models toward sound judgment.
Google added free textbooks to Gemini Notebook, letting students ground study guides and quizzes directly in course materials from OpenStax and S Chand.
Google's newsroom published a short guide showing how the Gemini app can turn rough, unedited notes about past jobs into polished resume bullets.
Google added a library of immersive explorations to the Gemini app's student hub, letting learners move through topics layer by layer from one interactive panel.
OpenAI highlighted how legal technology company Legora builds its agentic platform on ChatGPT, Codex, and OpenAI's models for law firms.
Nvidia detailed new RTX Spark laptop updates, including a transcription model from its Nemotron line and support for Liquid AI's LFM3 models.
Anthropic struck a new agreement with the US General Services Administration that gives government buyers a 40% discount on Claude Enterprise, Claude for Government and the Claude API.
A new preprint describes AdaptTok-Indic, a tokenization method that cuts token fertility for Bengali and Hindi text compared with standard subword tokenizers.
OpenAI is widening its partnership with Oracle so Oracle's business applications and AI agents appear directly inside ChatGPT for enterprise users.
Anthropic's pricing page now lists Claude's newest models, Opus 5.5, Sonnet 5.5 and Haiku 5.5, next to renamed Claude Cowork and unchanged Pro and Max plan rates.
OpenAI released ChatGPT Signals data tracking 165 work activities across nearly 100 countries, showing how work use of the assistant changes over time.
This is a permanent listing in what changed in AI. Each item keeps its own page, source and dates. How the stream is made.