The Daily Newsletter for Intellectually Curious Readers

If you're frustrated by one-sided reporting, our 5-minute newsletter is the missing piece. We sift through 100+ sources to bring you comprehensive, unbiased news—free from political agendas. Stay informed with factual coverage on the topics that matter.

Join for free today!

Today:

• OpenAI 100x Bigger GPT

• Sutskever’s SSI Raises $1 Billion

• GroqCloud Unveils LLaVA v1.5 7B

• AlphaProteo Advances Health Research with AI

• HyperWrite Launched Powerful Reflection 70B

• Claude Enterprise Challenges OpenAI’s ChatGPT

• OpenAI Reaches 1M Paid Users

BREAKING!! OpenAI JUST Announced GPT-5 [100X BIGGER]

OpenAI's GPT-5 is set to release in 2024 and is projected to be 100 times more powerful than GPT-4. This increase in computational capacity reflects both hardware advancements and algorithmic improvements. 

The model is expected to significantly outperform its predecessor, continuing the trend of exponential growth in AI capabilities. The announcement, made by OpenAI Japan, aligns with previous hints from Microsoft about the model’s capabilities.

WATCH THE VIDEO ON YOUTUBE

OpenAI co-founder Sutskever's new safety-focused AI startup SSI raises $1 billion

OpenAI co-founder Ilya Sutskever has launched a new AI startup, Safe Superintelligence (SSI), which raised $1 billion to develop safe AI systems. Valued at $5 billion, SSI aims to acquire computing power and hire top talent, focusing on safety and foundational AI research. 

The company plans to build a trusted team split between Palo Alto and Tel Aviv. Major investors include Andreessen Horowitz and Sequoia Capital. Sutskever, alongside co-founders Daniel Gross and Daniel Levy, emphasizes a strong company culture and a unique approach to scaling AI, setting SSI apart from competitors like OpenAI.

REUTERS

Introducing LLaVA V1.5 7B on GroqCloud

GroqCloud has launched LLaVA v1.5 7B, a multimodal AI model integrating image, audio, and text, on its Developer Console. LLaVA (Large Language and Vision Assistant) combines OpenAI’s CLIP and Meta’s Llama 2 7B, excelling in tasks like visual question answering, caption generation, and multimodal dialogue. 

Trained in 2023, it achieved top performance across seven benchmarks, highlighting its advanced visual reasoning. LLaVA’s applications span industries—retail inventory tracking, customer service chatbots, and educational tools. Now available in Preview Mode, LLaVA v1.5 7B empowers developers to innovate with multimodal AI on GroqCloud.

GROQ

AlphaProteo generates novel proteins for biology and health research

Google DeepMind has introduced AlphaProteo, an AI system that designs novel protein binders, enhancing drug design, disease research, and more. Unlike traditional methods that are time-consuming and require extensive lab work, AlphaProteo creates proteins that bind to specific targets, like cancer-related VEGF-A, achieving higher success rates and stronger binding affinities than existing techniques. 

Validated by external research groups, AlphaProteo’s binders show practical biological function, such as preventing SARS-CoV-2 infection. Though successful on multiple targets, AlphaProteo faces challenges with complex proteins like TNFɑ. DeepMind is committed to expanding AlphaProteo’s capabilities while maintaining a responsible approach to biosecurity.

GOOGLE DEEPMIND

Meet the new, most powerful open source AI model in the world: HyperWrite’s Reflection 70B

HyperWrite has launched Reflection 70B, an open-source AI model based on Meta’s Llama 3.1-70B, which introduces a unique error self-correction technique, making it the most powerful open-source AI model. Reflection 70B outperforms existing models by identifying and correcting errors in its outputs, enhancing accuracy for complex tasks. 

Developed in partnership with Glaive, which provided high-quality synthetic data for training, the model is part of a broader Reflection series, with an even larger Reflection 405B model expected soon. 

VENTUREBEAT

Anthropic launches Claude Enterprise plan to compete with OpenAI

Anthropic has launched Claude Enterprise, a subscription plan for its AI chatbot designed for businesses seeking more control and security. Competing with OpenAI's ChatGPT Enterprise, Claude Enterprise allows companies to upload proprietary data, enabling Claude to analyze, answer questions, and assist with tasks like creating graphics. 

It features a 500,000-token context window, GitHub integration, and unique workspaces for collaboration. Unlike competitors, Anthropic does not train on customer data. Early adopters include GitLab and Midjourney. 

TECHCRUNCH

OpenAI Hits 1 Million Paid Users For Business Versions of ChatGPT

OpenAI has reached over 1 million paid users for its corporate ChatGPT versions, including ChatGPT Team, Enterprise, and Edu. This milestone highlights growing demand from businesses and educational institutions despite competition. 

ChatGPT Enterprise, launched a year ago, offers enhanced features and privacy for companies, while ChatGPT Team targets smaller businesses. Nearly half of OpenAI’s corporate users are in the US, with strong adoption in Germany, Japan, and the UK. The surge in paid users reflects OpenAI’s strategy to generate revenue and manage the high costs associated with AI development.

BLOOMBERG

🧠RESEARCH

Loopy: Taming Audio-Driven Portrait Avatar with Long-Term Motion Dependency

Loopy is a new audio-driven video generation model that enhances natural motion in portraits without spatial constraints. By using advanced temporal and audio-to-latents modules, it learns long-term motion patterns, improving lifelike motion and synchronization with audio. Experiments show Loopy surpasses existing methods, delivering high-quality, realistic results.

Kvasir-VQA: A Text-Image Pair GI Tract Dataset

Kvasir-VQA is a specialized dataset combining 6,500 GI tract images with question-and-answer annotations, derived from HyperKvasir and Kvasir-Instrument datasets. It supports advanced machine learning tasks like image captioning, Visual Question Answering (VQA), and synthetic image generation, enhancing GI diagnostics and medical image analysis. The dataset is available for public use.

LongLLaVA: Scaling Multi-modal LLMs to 1000 Images Efficiently via Hybrid Architecture

LongLLaVA is a hybrid multi-modal large language model (MLLM) designed to handle up to 1,000 images efficiently. It combines Mamba and Transformer blocks, optimizing data construction and training strategies to improve performance and reduce computational costs. The model excels in long-context tasks like retrieval and ordering, offering high throughput and low memory usage.

OLMoE: Open Mixture-of-Experts Language Models

OLMoE is an open, cutting-edge language model using sparse Mixture-of-Experts (MoE) with 7 billion parameters but only 1 billion active per input token. Pretrained on 5 trillion tokens, it outperforms models with similar active parameters and even larger ones. OLMoE's model weights, training data, code, and logs are fully open-sourced.

LongCite: Enabling LLMs to Generate Fine-grained Citations in Long-context QA

LongCite enhances long-context language models by generating fine-grained, sentence-level citations, improving response accuracy and trustworthiness. It introduces LongBench-Cite to evaluate citation quality and develops CoF, a pipeline for precise QA generation. Models trained on the LongCite-45k dataset achieve state-of-the-art citation quality, surpassing advanced models like GPT-4o.

🛠️TOP TOOLS

Anthropic Quickstarts - A collection of projects designed to help developers quickly get started with building deployable applications using the Anthropic API

You.com - Leverage a personal AI search agent and customized recommendations with AI chatbot.

Intercom - The entire Intercom platform is powered by AI—so customers get instant support with an AI agent, agents get instant answers with an AI copilot, and support leaders get instant AI insights.

Open Funnel - End to end solutions to organize, manage and generate developer leads in the most Developer-Centric Way

Pinecone - Lets you deliver remarkable GenAI applications faster.

📲SOCIAL MEDIA

🗞️MORE NEWS

US, Britain, EU to sign first international AI treaty

The first international treaty on AI, signed by the U.S., Britain, and the EU, addresses AI's impact on human rights. Despite being a significant step, the treaty has been criticized for its vague principles and double standards, particularly in its limited scrutiny of private companies and national security exemptions.

OpenAI Considers Higher Priced Subscriptions to its Chatbot AI

OpenAI is considering high-priced subscriptions for its new AI models, including the reasoning-focused Strawberry and flagship model Orion, with potential monthly fees reaching up to $2,000. The pricing discussions are ongoing, and the final subscription costs are expected to significantly impact the competitive landscape among AI providers like Google and Anthropic.

Nvidia gets subpoena from US DoJ, Bloomberg News reports

The U.S. Department of Justice has subpoenaed Nvidia as part of an antitrust investigation into the company's market practices, including concerns over limiting supplier options and penalizing buyers not exclusively using Nvidia’s AI chips. This scrutiny comes amid Nvidia’s declining stock performance and increasing regulatory inquiries from various global authorities.

Paradigm launches to reinvent the spreadsheet with generative AI, filling in 500 cells per minute

Paradigm, a startup backed by Y Combinator, is revolutionizing spreadsheets with generative AI, automating data entry at 500 cells per minute. Founded by 22-year-old Anna Monaco, Paradigm uses AI agents to gather, analyze, and populate data, aiming to drastically improve productivity for data-heavy industries like consulting and recruiting.

YouTube is making new tools to protect creators from AI copycats

YouTube is developing tools to protect creators from AI-generated copycats, including technology to detect synthetic singing and facial deepfakes. Integrated with its Content ID system, these tools aim to safeguard creators’ voices and likenesses while promoting responsible AI use. YouTube also plans to enhance controls over third-party AI access to creator content.

Medical startup Paige unveils Alba AI all-in-one assistant for pathology research

Medical startup Paige has launched Alba AI, an all-in-one assistant designed to assist pathologists by consolidating patient data from various sources, streamlining diagnostics, and reducing administrative tasks. Currently for research use only, Alba enhances decision-making by summarizing medical records, generating reports, and integrating clinical-grade AI tools to aid cancer diagnosis, leveraging Paige's extensive AI models and vast pathology datasets.

Salesforce to acquire AI voice agent firm Tenyx, joining AI talent race

Salesforce is set to acquire Tenyx, an AI-powered voice agent startup, to enhance its AI-driven solutions. Tenyx, founded in 2022, serves various industries, including e-commerce and healthcare. The acquisition aligns Salesforce with tech giants like Microsoft and Amazon, which are also competing for AI talent to boost their AI capabilities.

Sakana AI scores $100M to challenge OpenAI, Anthropic as ‘world class’ AI lab

Sakana AI, a Japanese startup founded by former Google researchers, has raised $100 million to challenge leading AI labs like OpenAI and Anthropic. Backed by Nvidia and top investors, Sakana uses a unique, nature-inspired approach to develop high-performing AI models. The funding will help Sakana scale its infrastructure, expand its team, and build a world-class AI research lab in Japan.

Musician Indicted Over Years-Long Streaming Fraud Scheme After Allegedly Making Over $10 Million on AI Tracks

A North Carolina musician, Michael Smith, has been indicted for allegedly orchestrating a $10 million streaming fraud scheme using AI-generated music. Smith and his co-conspirators used bot accounts and exploited multi-account Family plans on platforms like Spotify and Apple Music to inflate streams, generating fraudulent royalties. The DOJ charged him with wire fraud and money laundering conspiracy.

Leaked messages show early challenges for Amazon's big AI product and concern about losing customers to Microsoft

Amazon's AI product, Q, is struggling post-launch due to high costs, missing features, and integration issues, leading to customer loss to Microsoft’s Copilot. Internal feedback highlights rushed development and unmet expectations, with some employees warning of further problems. Despite challenges, Amazon remains optimistic, citing strong growth and ongoing feature updates for Q.

AI breakthrough raises hopes for better cancer diagnosis

A new AI model 'Chief' can accurately detect multiple cancer types, assess treatments, and predict survival rates. The model outperforms other AI methods by up to 36% in cancer detection and treatment response. AI advancements in medical diagnosis show promise in providing accurate insights, improving diagnosis accuracy, and prognosis.