Last Week in AI #260: Figure raises $675 million, Microsoft partners with Mistral, Elon Musk sues OpenAI, India mandates Gen AI approvals
OpenAI, Nvidia, Bezos and others invest in Figure, Microsoft invests in Mistral, Elon Musk challenges OpenAI's nonprofit status, India requires Gen AI to be government approved before use
Top News
Figure rides the humanoid robot hype wave to $2.6B valuation
Figure, a Bay Area-based robotics firm, has raised a staggering $675 million in a Series B funding round, bringing its valuation to $2.6 billion. The investment round attracted a host of heavyweight backers, including Microsoft, OpenAI Startup Fund, Nvidia, Amazon Industrial Innovation Fund, Jeff Bezos through Bezos Expeditions, and several others, marking a significant vote of confidence from the tech and investment community. The company, which was initially bootstrapped by founder Brett Adcock with $100 million, aims to revolutionize the field of humanoid robots, having already achieved its ambitious goal of creating a walking bipedal robot within a year of its launch in 2022. Figure's strategy focuses on leveraging the latest advancements in AI and robotics to develop humanoid robots capable of performing a variety of tasks, particularly in industrial settings such as warehouses and factories. This new funding will support Figure's go-to-market strategy, including a partnership with BMW for robotics deployment, as well as a collaboration with OpenAI to enhance the AI capabilities of its robots for more natural human-robot interaction.
Microsoft made a $16M investment in Mistral AI
Microsoft has invested €15 million ($16.3 million) in Paris-based AI startup Mistral AI, as part of a partnership that also includes a distribution deal for Azure, Microsoft's cloud computing service. This investment, which is an extension of Mistral AI's Series A funding round, does not alter the startup's valuation, currently at about $2 billion following a significant funding round in December 2023. Mistral AI recently unveiled Mistral Large, a proprietary large language model aimed at competing with leading models like OpenAI's GPT-4. Unlike its previous models, Mistral Large is not open source and is accessible through Mistral’s API platform. This collaboration not only signifies Microsoft's expanding interest in AI technologies, having invested in both OpenAI and now Mistral AI, but also positions Mistral AI to potentially attract more customers through Azure. The investment has drawn the attention of the European Commission, which is scrutinizing the deal as part of its broader examination of relationships between large tech companies and generative AI firms, highlighting the growing concerns over competition and market power in the AI sector.
OpenAI says in memo that Musk’s claims ‘stem from Elon’s regrets’ that he’s not part of company
OpenAI executives have countered Elon Musk's allegations in a recent lawsuit, suggesting that Musk's actions are driven by his regret over no longer being involved with the AI startup he co-founded. In an internal memo, OpenAI's Chief Strategy Officer Jason Kwon expressed disappointment in Musk's lawsuit against the company, highlighting Musk's significant early contributions, including over $44 million in funding and providing the initial office space. The memo acknowledges Musk's foundational role but also notes disagreements over the direction of the company, including Musk's desire for full control and a proposed merger with Tesla, which OpenAI rejected to stay true to its mission. Despite Musk's critique of OpenAI's partnership with Microsoft and its shift towards a for-profit model, the company maintains its commitment to benefiting humanity. The conflict marks a significant public fallout between Musk and the AI organization he helped establish, amidst OpenAI's growing prominence following the release of the ChatGPT chatbot.
Indian government asks AI platforms to seek approval for deploying under-trial AI; makes labelling mandatory
In response to a controversy sparked by Google's AI platform's comments on Prime Minister Narendra Modi, the Indian government has mandated that social media and other platforms label under-trial AI models and prevent the dissemination of unlawful content. The Ministry of Electronics and Information Technology, on March 1, issued an advisory to intermediaries/platforms, cautioning that non-compliance could lead to criminal action. Platforms are required to ensure that their AI models or algorithms do not allow users to share illegal content, holding them accountable for any breaches. The advisory emphasizes the need for platforms to disclose and seek consumer consent for deploying under-trial AI models, labeling them to indicate potential unreliability. It suggests using a "consent popup" mechanism to inform users about the fallibility of AI outputs and mandates government approval before deploying such AI models. This directive builds on a previous advisory from December 26, 2023, addressing deepfakes and misinformation, underlining the government's intent to ensure a safe and trustworthy digital environment without taking direct control.
Other News
Tools
Mistral AI releases new model to rival GPT-4 and its own chat assistant - Mistral AI releases new model to rival GPT-4 and its own chat assistant, as well as announcing a partnership with Microsoft to provide Mistral models to its Azure customers.
Lightricks announces AI-powered filmmaking studio to help creators visualize stories - Lightricks has announced a new AI-powered filmmaking tool called LTX Studio, which helps creators from the ideation phase to generate an AI-powered short clip to understand how a storyline would play out, and the company is consolidating its products to focus on AI for both consumers and professionals.
Qualcomm inserts GenAI into smartphones at industry's mega tradeshow - Qualcomm is integrating AI into smartphones and other devices, showcasing a large language model running on an Android phone, an AI Hub for developers, and AI-infused 5G modem and Wi-Fi 7 silicon at MWC.
Lenovo debuts AI PCs that have specs a lot like vanilla PCs with this year's accelerated CPUs - Lenovo debuts AI PCs with the latest CPUs and NPUs, showcasing updated ThinkPads and a concept transparent display laptop, while also introducing a multi-cloud edge architecture for AI tasks.
Microsoft’s Windows 11 Copilot gets smarter with new plugins and skills - Microsoft's Windows 11 Copilot is getting smarter with new plugins and skills, allowing it to handle more everyday tasks and integrate with various services, while also adding AI editing integrations and improvements to widgets and Windows snap functionality.
Microsoft’s GitHub Offers Companies Souped-Up AI Coding Tool - GitHub releases a pricier paid version of its AI coding tool, Copilot Enterprise, designed to help engineers work faster and resolve issues more easily.
Adobe previews new cutting-edge generative AI tools for crafting and editing custom audio - Adobe is developing new generative AI tools for crafting and editing custom audio, allowing users to generate music from text prompts and then have fine-grained control to edit the audio for their precise needs.
Qualcomm's new AI Hub is a dream tool for developers building on-device models - Qualcomm unveils the AI Hub, a library of over 75 generative AI models that developers can easily download onto Qualcomm-powered devices, with hardware-aware optimizations for superior on-device AI performance.
Vimeo's new AI hub vows to organize your team's videos every which way - Vimeo introduces AI-powered hub for business teams to centralize and optimize video communications, offering features such as video library integration, transcript-based search, and AI-generated text summaries and highlights.
Copilot for OneDrive will fetch your files and summarize them - Microsoft's Copilot for OneDrive will use AI to find, summarize, and extract information from a variety of files, respond to natural language prompts, and create outlines, tables, and lists for users.
Windows just got its own Magic Eraser to AI-modify your photos - Windows PCs now have a new AI feature called Generative erase in the Photos app, allowing users to selectively remove objects and people from their photos.
Ideogram: Free AI image generator rolls out text-in-image upgrade - Ideogram 1.0 is an advanced AI image generator that can create legible and relevant text within images, offering a new subscription model and Magic Prompt feature to help users fine-tune their prompts.
Google brings Stack Overflow’s knowledge base to Gemini for Google Cloud - Google partners with Stack Overflow to integrate its knowledge base into Gemini for Google Cloud, providing AI companies with access to validated answers and aiming to merge human and AI expertise for developers.
The Humane AI Pin worked better than I expected — until it didn’t - The Humane AI Pin is a cool gadget with impressive features, but it is burdened by marketing and practical limitations, leading to doubts about its intended purpose and value.
Business
Startup Ideogram Raises $80 Million for AI Image Generation - Startup Ideogram secures $80 million in funding for its AI image generation technology from investors including Andreessen Horowitz and Index Ventures.
AI startup Glean lures Citigroup as investor at $2.2 billion valuation after revenue quadruples - AI startup Glean secures $200 million investment from Citigroup and others, with a $2.2 billion valuation, as its revenue quadruples and plans to expand into various industries.
Deutsche Telekom showcases futuristic AI-powered app-less smartphone - Deutsche Telekom showcases a concept AI-powered smartphone that relies entirely on AI, removing traditional phone features and predicting a future where apps will be obsolete.
SK Telecom partners with AI search startup Perplexity in Korea - SK Telecom partners with U.S. AI startup Perplexity to offer an alternative AI-based search engine to its users, potentially leading to further financial expansion and joint ventures.
AI chip startup Groq forms new business unit, acquires Definitive Intelligence - Groq, a startup developing AI chips, forms a new business unit, Groq Systems, and acquires Definitive Intelligence to expand its customer base and developer ecosystem, aiming to make AI accessible and affordable for all.
Meta unveils team to combat disinformation and AI harms in EU elections - Meta unveils a dedicated team to combat disinformation and AI harms in EU elections, bringing together experts to tackle misinformation, influence operations, and risks related to the abuse of AI.
Google’s Gemini AI picture generator to relaunch in a 'few weeks' following mounting criticism of inaccurate images - Google plans to relaunch its AI image generator Gemini after criticism of inaccurate images, which has reignited a debate within the AI industry about ethics and the company's commitment to AI.
Alphabet Drops During Renewed Fears About Google’s AI Offerings - Alphabet Inc. stock falls amid renewed fears over Google's AI missteps, with a 4.4% drop following the pause of a controversial image generation feature.
FlowGPT is the wild west of GenAI apps - FlowGPT is a platform that aims to be an "app store" for generative AI models, allowing users to build and share GenAI-powered apps, but it faces challenges with moderation and ethical concerns.
OpenAI says in memo that Musk's claims ‘stem from Elon’s regrets’ that he’s not part of company - Elon Musk's lawsuit against OpenAI is disputed by the company, which claims that Musk's regrets about not being involved with the company today may be the reason behind the legal action.
Inkitt, the self-publishing platform using AI to develop bestsellers, nabs $37M - Inkitt, a self-publishing platform, uses AI to develop bestsellers and aims to build a multimedia empire around personalized content, attracting significant funding and challenging traditional publishing.
Google Is Paying Publishers to Test an Unreleased Gen AI Platform - Google is paying independent publishers to test an unreleased generative artificial intelligence platform, which allows them to produce content using public data sources and potentially threatens the commercial foundations of digital publishing.
Apple Investors Reject Call for Report Into Company’s AI Use - Apple investors reject labor-backed proposal for AI transparency report, raising concerns about ethical use of technology.
It’s official: Waymo robotaxis are now free to use freeways and leave San Francisco - Waymo's self-driving car service has received approval to expand into Los Angeles and San Mateo counties, allowing it to use freeways to transport passengers, although the timeline for the expansion remains uncertain.
Research
Genie: Generative Interactive Environments - A new generative interactive environment, Genie, has been developed from unlabelled Internet videos, allowing users to prompt the model to create action-controllable virtual worlds through text, images, and sketches.
New AI model could streamline operations in a robotic warehouse - AI model uses deep learning to streamline operations in robotic warehouses, improving efficiency by dividing robots into groups and predicting the best areas to decongest, with potential applications in other complex planning tasks.
MegaScale: Scaling Large Language Model Training to More Than 10,000 GPUs - Training large language models at the scale of more than 10,000 GPUs presents unprecedented challenges to training efficiency and stability, requiring a full-stack approach that co-designs algorithmic and system components.
ChatMusician: Understanding and Generating Music Intrinsically with LLM - A new open-source LLM called ChatMusician is introduced, which can understand and generate music using ABC notation as a second language, achieving impressive results surpassing GPT-4 and GPT-3.5 on music understanding benchmarks.
Nemotron-4 15B Technical Report - Introducing Nemotron-4 15B, a 15-billion-parameter multilingual language model demonstrating strong performance in English, multilingual, and coding tasks.
MobileLLM: Optimizing Sub-billion Parameter Language Models for On-Device Use Cases - Efficient sub-billion parameter language models for mobile devices are designed, emphasizing the significance of model architecture over data and parameter quantity, resulting in improved accuracy and performance.
EMO: Emote Portrait Alive - Generating Expressive Portrait Videos with Audio2Video Diffusion Model under Weak Conditions - A novel framework called EMO utilizes a direct audio-to-video synthesis approach to generate highly expressive and lifelike talking head videos, outperforming existing methodologies in terms of expressiveness and realism.
Griffin: Mixing Gated Linear Recurrences with Local Attention for Efficient Language Models - A proposal for efficient language models using a hybrid model that combines gated linear recurrences with local attention, achieving high performance and hardware efficiency.
VastGaussian: Vast 3D Gaussians for Large Scene Reconstruction - A new method called VastGaussian uses 3D Gaussian Splatting to achieve high-quality reconstruction and real-time rendering on large scenes, outperforming existing NeRF-based methods.
How to think step-by-step: A mechanistic understanding of chain-of-thought reasoning - Understanding chain-of-thought reasoning through a mechanistic approach.
RNNs are not Transformers (Yet): The Key Bottleneck on In-context Retrieval - RNNs and Transformers face a key bottleneck in in-context retrieval, impacting their AI capabilities.
Concerns
AI Warfare Is Already Here - AI is being used in warfare by the US military to identify and strike targets, with the Pentagon investing billions in AI-related activities and the National Geospatial-Intelligence Agency leading the development of the Maven platform, despite concerns about the accuracy and ethical implications of AI decision-making in combat.
AI ‘dream girls’ are coming for porn stars’ jobs - AI is revolutionizing the adult entertainment industry, with companies racing to create custom AI models for creating photorealistic images and videos, raising concerns about compensation for performers and the potential for abuse.
Police investigate AI-generated nude photos of students in Beverly Hills - AI-generated nude photos of students at a middle school in Beverly Hills are being investigated by the police, raising concerns about the proliferation of nonconsensual sexually explicit deepfakes targeting women and girls.
‘Road House’ brawl: Amazon used AI to replicate actors’ voices during strike, lawsuit alleges - Amazon Studios is being sued for allegedly using AI to replicate actors' voices in the 2024 remake of "Road House" in an attempt to finish the movie before the copyright expired.
Hugging Face, the GitHub of AI, hosted code that backdoored user devices - AI models uploaded to Hugging Face were found to contain hidden and unwanted actions, including a model that opened a reverse shell, giving a remote device full control over the end user’s device.
OpenAI accuses NYT of hacking ChatGPT to set up copyright suit - OpenAI accuses The New York Times of hacking ChatGPT to set up a copyright suit, claiming that the Times paid someone to exploit a bug in OpenAI's models to generate fake NYT content and gather evidence for their claims.
The Intercept, Raw Story, and AlterNet sue OpenAI and Microsoft - Three news organizations sue OpenAI and Microsoft for alleged copyright infringement, claiming that AI models trained by the companies reproduce copyrighted works without providing author, title, or copyright information.
Man admits to paying magician $150 to create anti-Biden robocall - Political consultant admits to paying magician to create fake anti-Biden robocall using AI, sparking investigation and leading to immediate FCC regulation.
Copyright © 2024 Skynet Today, All rights reserved.