Explosive OpenAI, Musk emails revealed

Welcome, AI enthusiasts.

Latest court documents just unearthed a trove of emails between OpenAI and Elon Musk, giving us a fly-on-the-wall take a look at the early days of the world’s most explosive AI drama.

From AGI dictatorship worries to million-dollar talent wars, the seeds of today’s AI power struggles were planted deeper than anyone knew. Let’s get into it…

In today’s AI rundown:

• Early emails between OpenAI, Musk revealed

• Suleyman: AI with ‘near-infinite’ memory achieved

• Transform blog posts into infographics

• Arc Institute releases ‘ChatGPT for DNA’

• 5 recent AI tools & 5 recent AI jobs

• More AI & tech news

Read time: 4 minutes

Friendly reminder: We’re moving to a brand new email sending address. Make sure that so as to add [email protected] to your contacts to make sure you keep receiving our emails.

LATEST DEVELOPMENTS

OPENAI & ELON MUSK

📧 Early emails between OpenAI, Musk revealed

The Rundown: Latest court filings from Elon Musk’s lawsuit against OpenAI have exposed internal emails from the corporate’s early life — revealing intense early tensions around control, talent acquisition, and concerns about AI dominance.

The small print:

• The newly released emails are dated between 2015 and 2018, covering OpenAI’s initial formation to Musk’s eventual departure from the corporate.

• Leadership anxious about control over AGI, with Ilya Sutskever warning of concerns about potential “dictatorship” scenarios from Google or internally.

• Early salary battles erupted when DeepMind attempted to poach OpenAI’s founding team, forcing rapid compensation increases of $100-200K per person.

• Internal debates around Microsoft collaboration surfaced, with Musk stating he’d pay “$50M to not look like Microsoft’s marketing b*tch.”

• Altman’s role also sparked concerns, with Sutskever questioning his ‘cost function’ and whether AGI was his primary motivation.

Why it matters: At any time when each Elon and Sama are involved, you understand the drama might be juicy. These email exchanges are incredible artifacts from the early days of OpenAI’s formation and shed additional light on Musk’s vendetta against Altman and the undercurrents which will have contributed to the board ouster chaos in Nov. 2023.

TOGETHER WITH ASSEMBLY AI

🗣️ Capture the complexity of human speech

The Rundown: Universal-2 by AssemblyAI offers advanced speech-to-text capabilities, transforming audio into precise and readable transcripts that capture the complexity of human speech beyond traditional metrics.

With Universal-2, you’ll experience:

• A 15% improvement in text formatting, preserving dates, times, and costs

• Greater accuracy for correct nouns and alphanumeric data

• Enhanced conversation intelligence for sharper insights and faster workflows

Able to experience Universal-2’s next-gen accuracy? Start at no cost.

MICROSOFT

🧠 Suleyman: AI with ‘near-infinite’ memory achieved

The Rundown: Microsoft AI CEO Mustafa Suleyman just revealed the corporate has created prototypes with “near-infinite memory” capabilities in a brand new interview with Times Techies, calling it the ‘critical piece’ of AI development.

The small print:

• Microsoft’s prototypes can allegedly maintain persistent memory across unlimited sessions, breaking through current limitations.

• Suleyman expects this technology to be available by 2025, enabling AI systems that “just remember” with ongoing, evolving dialogues.

• Suleyman also said that memory is an ‘inflection point’ that makes it price investing time in chats, changing the present frustrating and shallow experience.

• The Microsoft AI CEO also noted a coming shift from AI understanding and seeing context to a real proactive companion over a reactive chatbot.

Why it matters: While we’ve seen memory efforts from systems like ChatGPT, Suleyman’s ‘hole’ description accurately portrays those early iterations. Unlocking the power for limitless memory can result in models that may form lasting, evolving relationships with users and higher understand their needs and goals.

AI TRAINING

📊 Transform blog posts into infographics

The Rundown: Infography uses AI to routinely turn any blog post or text content into professionally designed infographics.

Step-by-step:

1. Create a free account at Infography (you get 5 credits to begin).

2. Select your content source (URL, topic, or direct text).

3. Select your selected format (vertical, square, or presentation).

4. Edit and customize your design with brand colours and fonts.

Pro tip: Use the Brand section to avoid wasting your style preferences for consistent designs across all of your infographics.

PRESENTED BY CODEIUM

⚙️ Unlock true flow state with the Windsurf Editor

The Rundown: Windsurf Editor introduces the primary agentic IDE, integrating Codeium’s beloved extensions with Cascade — collaborative AI agents that merge copilot and agent functionalities for an unparalleled coding journey.

With Windsurf Editor, you may:

• Experience a transformative flow state while coding with AI

• Collaborate seamlessly with AI agents that boost your productivity

• Benefit from the combined advantages of copilot and agent systems in your IDE

Try Windsurf Editor now and step into tomorrow’s IDE today. 

AI RESEARCH

🧬 Arc Institute releases ‘ChatGPT for DNA’

The Rundown: Scientists on the Arc Research Institute just introduced Evo, an AI model trained on 2.7M microbial genomes that may each interpret and generate genetic sequences with unprecedented accuracy.

The small print:

• Unlike traditional language models trained on text, Evo concurrently learns from DNA, RNA, and protein sequences.

• In early tests, Evo already designed working genetic editing tools and accurately predicted how DNA changes would affect bacteria.

• Evo can generate entirely recent genome-length sequences over 1M base pairs long, though they aren’t able to forming fully viable organisms yet.

• The researchers deliberately excluded human-affecting viral genomes from training for safety reasons.

Why it matters: Having a ‘ChatGPT for DNA’ is a mind-blowing unlock, with the potential to drastically speed up research, design recent proteins for drug development, predict disease-causing mutations, and more. Nonetheless, it also raises questions on how society will responsibly handle tools that may write genetic code as easily as we write emails.

NEW TOOLS & JOBS

Trending AI Tools

• ⚙️ Momen – Full-stack, no-code web app builder with an integrated AI agent development framework

• 🔉 Mikrotakt – AI-powered tool to extract vocals, instruments, drums, and more from any song

• 📊 Sharbo AI – Analyze, compare, and track competitor features relative to your product

• 💨 Sre.AI – AI DevOps agents for enterprise teams tailored to any workflow

• 🎲 AI Game Master – Fully immersive DnD-inspired experience powered by AI

Latest AI Job Opportunities

• 📱 The Rundown – Social Media Manager

• 🏗️ Curai – Senior Implementation Manager

• 🎨 Grammarly – Staff Growth Designer, Monetization

• 📊 Databricks – Director of Product Operations

• 💻 Snorkel – Senior IT Specialist (Contract)

QUICK HITS

Stanford researchers unveiled SEQUOIA, an AI system that may predict gene expression patterns in cancer cells by analyzing standard biopsy images, potentially eliminating the necessity for expensive testing.

Kai-Fu Lee’s 01.ai revealed a breakthrough in efficient AI training, achieving competitive results in comparison with OpenAI’s reported $1B investment into training GPT-5.

The MIT Jameel Clinic released Boltz-1, an open-source biomolecular model that matches Google DeepMind’s AlphaFold3’s accuracy in predicting 3D structures.

Nvidia’s upcoming Blackwell AI chips reportedly suffer overheating issues, prompting design revisions and raising concerns about data center deployment timelines.

Google’s Gemini AI chatbot sparked concerns after delivering a threatening message telling a Michigan student to ‘die’ during a routine homework help conversation, prompting the corporate to acknowledge a security filter failure.

U.S. President Joe Biden and China’s Xi Jinping reached recent landmark agreements on AI nuclear controls within the pair’s final meeting before the administration change, ensuring that only humans will make decisions with nuclear weapons.

Coca-Cola released a brand new AI-generated Christmas commercial, partnering with Silverside AI to reimagine its original “Holidays Are Coming” spot.

THAT’S A WRAP

See you soon,

Rowan, Joey, Zach, and Alvaro—aka The Rundown Team