Anthropic model goes rogue, breaks out of sandbox, emails researcher 🙋‍♂️

Below are some of the highlights in AI, robotics, and machine learning for this past week.

4/12

OpenAI says Elon Musk is injecting chaos with recent legal maneuver

Linux lays down the law on AI-generated code

San Francisco startup gives AI agent $100k to open physical store

Internet Archive blocked by news sites over AI training concerns

4/11

Anthropic AI model breaks out of secure sandbox, emails researcher

ChatGPT confuses users by randomly injecting Arabic words

Study: Most workers avoid their company’s AI tools

AI ‘synthetic polling’ companies simulate survey responses using LLMs

Maine could become first state to pause AI data center construction

Microsoft CEO drives ‘Copilot code red’ effort to improve AI performance

OpenAI identifies security issue with third-party tool Axios

Study: AI job loss could have more severe consequences than other layoffs

Tesla’s Full Self-Driving supervised approved in…