OpenAI Codex Launch, Claude Caught Faking Sources, and LLMs Struggle With Context

In today’s episode of AI Deep Dive, we explore the launch of OpenAI’s Codex, a powerful new agent aimed at enhancing software engineering via ChatGPT. But not all AI news is upbeat—Anthropic’s Claude has come under fire for citing fake legal information in a real case, highlighting serious trust issues. We also dig into new research showing large language models struggle with multi-turn conversations, and examine how Cognichip is using AI to accelerate chip design. Don’t miss this deep dive into the cutting edge—and pitfalls—of today’s AI tools.