Anthropic researcher quits, citing AI safety fears

A Reddit user describes a workflow of discussing a plan with Claude chat, then asking it to generate Claude Code prompts, and pasting the output back into chat for iteration. The post asks if this approach is effective.

Infercat is a new tool that lets users share their local AI models with friends over an encrypted peer-to-peer tunnel, using invite codes for access from anywhere. It aims to make local AI as convenient as cloud-based services.

OpenAI used an unreleased model to solve the Navier-Stokes Millennium Prize Problem, but NYU professor Tristan Buckmaster accused the company of using his and Anthropic's Levent Alpöge's work without permission. OpenAI spent $22.5M on the effort and admitted it "cannot rule out" that de-identified data from their usage helped improve models.

Zepto, one of India's fastest-growing companies, uses Databricks and MLflow to build evaluation-first AI agents for real-time customer support. The approach emphasizes continuous evaluation to ensure reliability and scale.

US security agencies accused China's top AI companies, including DeepSeek and Moonshot AI, of systematically extracting proprietary knowledge from American firms, and warned Silicon Valley developers to protect their work.