#223 - Haiku 4.5, OpenAI DevDay, Claude Skills, Scaling RL, SB 243 - Last Week in AI Recap

Podcast: Last Week in AI

Published: 2025-10-24

Duration: 1 hr 12 min

Guests: Eric Schlantz

Summary

The episode delves into recent advancements in AI, covering updates from prominent companies like Anthropic, OpenAI, and Google. It also addresses business deals and policy changes impacting the AI landscape.

What Happened

Anthropic announced the release of Haiku 4.5, their smallest and fastest AI model, which performs impressively on benchmarks and is cost-effective for businesses. This smaller model appears to punch above its weight class, offering a cheaper alternative while maintaining high performance.

OpenAI's Dev Day introduced several new tools, including an app SDK that allows developers to embed applications within ChatGPT. This move aims to make ChatGPT a central interface for various services, enhancing its utility for both consumers and developers.

Anthropic also launched Claude Skills, which are modular capabilities that extend Claude's functionality. These skills allow businesses to create customized workflows, improving operational efficiency by packaging instructions and resources for repeated use.

Microsoft has integrated its Vibe working feature into Excel and Word, allowing AI to generate complex spreadsheets and documents. This feature is designed to enhance productivity by automating tasks that would typically require significant manual effort.

Google's release of VO 3.1 offers improved control over video generation, allowing users to choose different generation lengths and formats. This update is part of Google's strategy to cater to YouTube creators and enhance content creation tools.

In business, OpenAI has been making substantial deals with companies like Broadcom and AMD to expand its compute capacity, signaling its commitment to scaling AI infrastructure. These deals highlight the growing demand for AI capabilities and the competitive landscape among AI providers.

California has become the first state to regulate AI companion chatbots with the signing of SB 243. This law aims to protect children and vulnerable users from potential harms associated with AI chatbots, setting a precedent for future regulatory measures.

Key Insights