Good morning. It’s Friday, Jun 6th.

On this day in tech history: 1995: MySQL, an open-source database system was launched. Built on SQL, it provided fast, dependable data management in a client-server model. Its flexible design, with B-tree and hash table options and engines like MyISAM and InnoDB, made it a go-to for web apps like WordPress, thanks to its free GPL license.

In today’s email:

• Anthropic unlocks advanced Claude features

• Tech's New Power Play: From Lobbying to Governing AI

• 5 New AI Tools

• Latest AI Research Papers

You read. We listen. Let us know what you think by replying to this email.

Anthropic unlocks advanced Claude features, rejects 'blunt' AI moratorium in public stance

Anthropic is consolidating its position across commercial, enterprise, and governmental AI sectors with a series of feature rollouts, product expansions, and public policy stances. Claude Pro has been upgraded with three formerly premium-only tools. Claude Code for command-line tasks, third-party integrations for platforms like Jira and Asana, and a research module supporting in-depth analysis of web content, all now standard at no extra cost. To support long-context workflows, Claude’s “Projects” feature has also gained a retrieval mode that activates automatically when input limits are exceeded, extending memory across sessions.

Parallel to product expansion, Anthropic has launched Claude Gov, a model optimized for U.S. defense and intelligence use, engineered for secure handling of sensitive material under tailored usage exemptions. Amid growing regulatory scrutiny, CEO Dario Amodei has publicly opposed a proposed 10-year AI regulation moratorium, arguing it would stall both federal and state governance, and instead advocates for a unified national transparency standard. He also rebuffed concerns about Claude’s future following OpenAI’s acquisition of Windsurf, stating, “It would be odd for us to sell Claude to OpenAI,” reinforcing Anthropic’s independent trajectory. Read more.

Tech's New Power Play: From Lobbying to Governing AI

Major AI players are no longer just lobbying Washington, they’re moving in.

OpenAI, Meta, Anthropic, and xAI are forging deeper ties with government agencies, shifting AI’s focus from productivity and ethics to national security and geopolitical muscle. At the recent AI+ Expo, Eric Schmidt pushed for tighter public-private AI partnerships, while Anthropic rolled out defense-oriented AI models and Meta opened up its Llama tech to military partners. Commerce Secretary Howard Lutnick put it bluntly: winning the AI and quantum race is a top priority.

Meanwhile, The Trump-Musk rift has escalated into a full-blown policy crisis with lasting consequences for American tech. Trump’s administration, once reliant on Musk’s support, is now threatening to cut federal contracts and subsidies to his firms, citing his public criticism of the GOP’s research-slashing megabill. The fallout extends beyond personality politics: research funding is drying up, tariffs are hitting lab imports, and foreign scientists are being forced out. As AI firms latch onto defense deals and political clout, critics warn blurred industry-government lines risk eroding transparency, safety, and true oversight, putting innovation and accountability at stake. Read more. 

• X’s new policy prevents companies from using posts to ‘fine-tune or train’ AI models

• AI tool allows students to draw their handwriting and artwork in mid-air with their fingers

• DeviantArt’s co-founder made a digital art display... and it’s $22,000

• ElevenLabs Releases Alpha of Eleven v3, Enhanced Expressive Text-to-Speech Model

• Perplexity now has access to SEC data; received 780 million queries last month, CEO says

• Amazon launches new R&D group focused on agentic AI and robotics

• AMD’s Untether AI Deal Signals the Beginning of the End for GPU-Driven AI Training

• X Blocks AI Bots from Training on Its Data

• STRADVISION Partners with Arm to Drive the Future of AI-Defined Vehicles

• Dead Sea Scrolls mystery deepens as AI finds manuscripts to be much older than thought

• New AI Transforms Radiology With Speed, Accuracy Never Seen Before

• Artists used Google’s generative AI products to inspire an interactive sculpture

• Figure 02 fully autonomous driven by Helix learns to flip and flatten packages

• ‘String breaking’ observed in 2D quantum simulator

• The Oversight Board says Meta has an AI deepfake problem

• Higgsfield AI Launches Higgsfield Speak for Motion-Driven Talking Videos

• New Google DeepMind Podcast about Drug Discovery with the help of AI

• Breakthrough AI study identifies original authors of the Hebrew Bible

📄 Reflect, Retry, Reward: Self-Improving LLMs via Reinforcement Learning

📄 Advancing Multimodal Reasoning: From Optimized Cold Start to Staged Reinforcement Learning ​​

📄 An AI Capability Threshold for Rent-Funded Universal Basic Income in an AI-Automated Economy

📄 RoboRefer: Towards Spatial Referring with Reasoning in Vision-Language Models for Robotics

📄 ComfyUI-Copilot: An Intelligent Assistant for Automated Workflow Development

Thank you for reading today’s edition.

Your feedback is valuable. Respond to this email and tell us how you think we could add more value to this newsletter.

Interested in reaching smart readers like you? To become an AI Breakfast sponsor, reply to this email or DM us on 𝕏!