Sam Altman launches new eye-scanning orb for human verification
Plus Mistral AI launches new on-device models
Today’s Highlights:
📰 News: Sam Altman’s Worldcoin has rebrands to "World" as it launches a new eye-scanning orb for human verification + Mistral AI launches new on-device models
💰 Funding: CoreWeave secures a $650M loan + Lightmatter raises $400M Series D
⚡️ Top News Stories:
1. Sam Altman’s Worldcoin has rebranded to "World" as it expands its mission beyond cryptocurrency, unveiling faster, mass-producible Orbs to scale human verification globally, with the goal of proving personhood online and redistributing AI benefits using blockchain technology. Here’s what was announced:
New Orb unveiled: A faster, more efficient iris-scanning Orb was introduced to scale World ID verification globally.
World ID 3.0: Enhanced with new privacy features and World ID Credentials, allowing users to verify identity with passports.
World App 3.0: The app adds Mini Apps for third-party integration and new wallet features like Vault and World Pay.
World Chain launch: A blockchain designed for human-focused activity and transactions is now live.
Here’s a recap of the event:
2. Mistral AI launches new Ministraux models, Ministral 3B and 8B, designed for edge computing and efficient, on-device use cases.
These models offer up to 128k context length, supporting tasks like function-calling and autonomous workflows with low latency and high efficiency.
Designed for local inference in privacy-focused applications, such as on-device translation and robotics, enhancing AI's versatility.
3. Nvidia unveils Llama-3.1-Nemotron-70B-Instruct which outperforms competitors like OpenAI’s GPT-4o and Anthropic’s Claude 3.5 Sonnet in multiple benchmarks.
4. Google's NotebookLM has introduced updates allowing users to guide AI-generated audio conversations based on specific topics in documents or videos, with a new "Customize" feature to reduce irrelevant content and hallucinations. Additionally, Google launched a NotebookLM Business pilot, offering early access, training, and support for organizations, with plans to expand features such as native mobile apps, more languages, and additional voices for audio summaries.
5. Boston Dynamics has partnered with Toyota Research Institute (TRI) to integrate AI-based large behavior models into the Atlas humanoid robot, aiming to enhance its ability to autonomously perform complex tasks, with a focus on developing a general-purpose humanoid capable of tackling diverse real-world challenges.
6. Dropbox has launched Dash for Business, an AI-powered universal search tool that helps teams manage, search, and organize content across connected apps, integrating with platforms like Google Drive, OneDrive, Notion, and Asana for streamlined, platform-agnostic searches.
7. Amazon Ads has launched an AI creative studio and Audio generator to simplify ad creation, enabling advertisers to quickly generate images, videos, and 30-second audio ads using generative AI, with features like an AI gallery for inspiration and the ability to produce high-quality ads from a single product image.
8. Google has rebuilt its shopping experience using AI to make product searches more personalized and relevant, offering AI-generated shopping insights that provide tailored briefs highlighting key considerations, product recommendations, and specific details for shoppers.
9. Google.org has launched $15M in AI training grants, including $10M to establish the Center for Federal AI to train government leaders in responsible AI use, and $5M to InnovateUS to expand AI training programs to over 100,000 state and local public sector workers, with a focus on addressing societal challenges through responsible AI upskilling.
10. The Open Source Initiative (OSI) has criticized Meta for labeling its Llama AI models as "open-source," arguing that it confuses users and dilutes the term, as the models don't fully meet open-source standards due to a lack of transparency and restrictions on use by Meta's competitors, despite allowing developers to download Llama's weights.
11. OpenAI has pledged to use its patents only defensively, but experts criticize the vagueness of what constitutes “defensive” use, raising concerns that its broad definition of harmful activities could be used against competitors or critics, especially as its IP strategy relies more on trade secrets than its relatively small patent portfolio.
12. Tensions have emerged in the Microsoft-OpenAI partnership as OpenAI seeks more funding and computing power amid significant financial pressures, with the startup expecting a $5 billion loss this year, while Microsoft reassesses its dependency on OpenAI.
13. X (formerly Twitter) has updated its privacy policy, to allow third-party companies to use user data for training AI models unless users opt out, as part of its plan to monetize data and license it to AI firms amidst financial challenges.
14. The New York Times has issued a cease and desist letter to Perplexity, accusing the AI company of violating copyright laws by using its researched and edited journalism for summaries without proper licensing, resulting in "unjust enrichment."
15. Amazon, Google, and Microsoft are investing in nuclear energy to meet AI’s high energy demands, with Microsoft reviving the Three Mile Island plant and Google and Amazon focusing on small modular reactors (SMRs), as AI-driven data centers require significantly more electricity than traditional tech businesses, prompting the shift to more reliable power sources.
16. The Biden administration is considering country-specific export caps on AI chips from companies like Nvidia and AMD, with a specific on countries like the UAE and Saudi Arabia due to concerns about potential misuse of AI tech as these nations invest heavily in AI data centers.
17. The U.S. Treasury saved over $4Bn in fiscal 2024 using AI-driven fraud detection, including $2.5Bn from identifying high-risk transactions and $1Bn from check fraud recovery, with AI models efficiently analyzing large datasets to detect suspicious financial patterns and prevent $500M in additional losses.
💰 Top Funding News:
1. CoreWeave, the cloud-computing AI startup, secured a $650M loan led by JPMorgan, Goldman Sachs, and Morgan Stanley to support its growth.
2. Lightmatter, which uses photonic chips to create high-speed optical interconnects, allowing hundreds of GPUs to work synchronously in AI data centers, raised a $400M Series D, led by T. Rowe Price Associates, w/ Fidelity Management & Research Company and GV.
3. Decagon, which develops AI-powered customer service agents that outperform competitors in handling customer queries, raised a $65M Series B, led by Bain Capital Ventures, w/ Elad Gil, Accel, and Bond Capital.
4. Galileo, which specializes in evaluating AI models to ensure they don't hallucinate, leak sensitive data, or make mistakes, raised a $45M Series B led by Scale Venture Partners, w/ Premji Invest, Citibank Ventures, Databricks Ventures, and Clem Delangue.
5. Xscape Photonics, which develops silicon photonics technology to create high-bandwidth interconnects for GPUs and AI chips in data centers, raised a $44M Series A, led by IAG Capital Partners, w/ Cisco and Nvidia's venture arms.
6. Port, which integrates LLMs into its DevOps platform, raised a $35M Series B, led by Accel, w/ Bessemer Venture Partners, Team8, and TLV Partners.
7. OroraTech, which uses AI to predict and monitor wildfire risks by processing thermal data from over 25 satellites, raised a $27M Series B, led by Korys, the European Circular Bioeconomy Fund (ECBF), and Bayern Kapital.