#210 - Claude 4, Google I/O 2025, OpenAI+io, Gemini Diffusion

Our 210th episode with a summary and discussion of last week's big AI news!

Recorded on 05/23/2025

Hosted by Andrey Kurenkov (twitter.com/andrey_kurenkov) and Jeremie Harris (twitter.com/jeremiecharris).

Feel free to email us your questions and feedback at contact@lastweekinai.comand/or hello@gladstone.ai

Read out our text newsletter and comment on the podcast at lastweekin.ai.

Join our Discord here! (discord.gg/nTyezGSKwP) discord.gg/nTyezGSKwP

In this episode:

• Google's Gemini diffusion technology showcases significant improvements in speed and efficiency for generating text, potentially revolutionizing the auto-regressive generation paradigm.

• Anthropic activates AI Safety Level 3 protections for Claude Opus 4, implementing robust measures such as bug bounties, synthetic jailbreak data, and preliminary egress bandwidth controls to mitigate bio-risk threats.

• OpenAI responds to the California Attorney General, refuting claims by the not-for-private-gain coalition and defending their controversial restructuring plans amidst ongoing criticism.

• Mistral delays the release of its Llama 4 Behemoth model due to training challenges, while Meta faces similar obstacles in rolling out its large-scale AI models, signaling difficulties in reaching frontier level performance.

Timestamps + Links:

• (00:00:00) Intro / Banter

• (00:01:43) News Preview

• Tools & Apps

(00:02:58) Anthropic’s new Claude 4 AI models can reason over many steps (techcrunch.com/2025/05/22/anthropics-new-claud…)(00:09:58) Google Unveils A.I. Chatbot, Signaling a New Era for Search (nytimes.com/2025/05/20/technology/personaltech…)(00:14:04) Google rolls out Project Mariner, its web-browsing AI agent (techcrunch.com/2025/05/20/google-rolls-out-pro…)(00:16:40) Veo 3 can generate videos — and soundtracks to go along with them (techcrunch.com/2025/05/20/googles-veo-3-can-ge…)(00:21:26) Imagen 4 is Google’s newest AI image generator (techcrunch.com/2025/05/20/imagen-4-is-googles-…)(00:23:15) Google Meet is getting real-time speech translation (techcrunch.com/2025/05/20/google-meet-is-getti…)(00:25:36) Google’s new Jules AI agent will help developers fix buggy code (theverge.com/2024/12/11/24318628/jules-google-…)(00:26:43) GitHub’s new AI coding agent can fix bugs for you (theverge.com/news/669339/github-ai-coding-agen…)(00:28:50) Mistral’s new Devstral model was designed for coding (techcrunch.com/2025/05/21/mistrals-new-devstra…)

• Applications & Business

(00:29:53) OpenAI Unites With Jony Ive in $6.5 Billion Deal to Create A.I. Devices (nytimes.com/2025/05/21/technology/openai-jony-…)(00:36:10) OpenAI’s planned data center in Abu Dhabi would be bigger than Monaco (techcrunch.com/2025/05/16/openais-planned-data…)(00:41:18) LM Arena, the organization behind popular AI leaderboards, lands $100M (techcrunch.com/2025/05/21/lm-arena-the-organiz…)(00:45:21) Nvidia CEO says next chip after H20 for China won't be from Hopper series (reuters.com/world/china/nvidia-ceo-says-next-c…)(00:46:39) Google’s Gemini AI app has 400M monthly active users (techcrunch.com/2025/05/20/googles-gemini-ai-ap…)(00:51:15) AI Servers: End demand intact, but rising gap between upstream build and system production (2025.5.18) (x.com/Jukanlosreve/status/1924245891130802544)Projects & Open Source

(00:53:46) Meta Is Delaying the Rollout of Its Flagship AI Model (wsj.com/tech/ai/meta-is-delaying-the-rollout-o…)

• Research & Advancements

(00:57:53) Gemini Diffusion (blog.google/technology/google-deepmind/gemini-…)(01:03:07) Chain-of-Model Learning for Language Model (arxiv.org/abs/2505.11820)(01:09:16) Seek in the Dark: Reasoning via Test-Time Instance-Level Policy Gradient in Latent Space (arxiv.org/abs/2505.13308v1)(01:15:38) Two Experts Are All You Need for Steering Thinking: Reinforcing Cognitive Effort in MoE Reasoning Models Without Additional Training (arxiv.org/abs/2505.14681)(01:20:16) Lessons from Defending Gemini Against Indirect Prompt Injections (storage.googleapis.com/deepmind-media/Security…)(01:23:35) How Fast Can Algorithms Advance Capabilities? (epoch.ai/gradient-updates/how-fast-can-algorit…)(01:30:20) Reinforcement Learning Finetunes Small Subnetworks in Large Language Models (arxiv.org/abs/2505.11711v1)

• Policy & Safety

• (01:31:12) Exclusive: What OpenAI Told California's Attorney General (obsolete.pub/p/exclusive-what-openai-told-cali…)

• (01:38:25) Activating AI Safety Level 3 Protections (anthropic.com/news/activating-asl3-protections)

See Privacy Policy at art19.com/privacy and California Privacy Notice at art19.com/privacy#do-not-sell-my-info.