OpenAI launches Sora, it’s new text-to-video AI model
Plus Google launches next version of its Gemini LLM
Today’s Highlights:
📰 News: OpenAI introduces it’s new text-to-video AI model + Google launches next version of its Gemini LLM
💰 Funding: Github launches an AI accelerator, Langchain raises a Series A, and former Salesforce CEO launches AI Agent startup w/ $110M in funding from Sequoia and Benchmark.
⚡️ Top News Stories:
1.OpenAI introduces Sora, a new text-to-video AI model capable of creating photorealistic videos up to 1min long.
Sora can generate complex scenes with multiple characters, specific motions, and detailed backgrounds and subjects.
The model understands physical object interactions, accurately interprets props, and generates characters with vibrant emotions.
Sora can also transform a still image into a video, fill in missing frames, or extend existing videos.
Despite its capabilities, Sora may struggle with simulating complex scene physics and cause-and-effect relationships accurately.
The development of text-to-video AI models like Sora marks a significant advancement from text-to-image generators, with companies like Runway and Google's Lumiere as competitors.
Currently, Sora is available to "red teamers" for harm and risk assessment and selected visual artists, designers, and filmmakers for feedback.
OpenAI's will add watermarks to generated creations, indicating a concern for the potential misuse of AI-generated content.
2. Google announces Gemini 1.5, the successor to its Gemini LLM
Gemini 1.5 Pro matches the performance of the high-end Gemini Ultra, outperforming Gemini 1.0 Pro in 87% of benchmark tests.
The model employs a "Mixture of Experts" technique for efficiency, running only part of the model for each query.
A significant feature of Gemini 1.5 is its enormous context window of 1 million tokens, allowing for complex queries and extensive information processing.
CEO Sundar Pichai highlited the model's potential for businesses, enabling detailed analysis of vast amounts of data in a single query.
Initially, Gemini 1.5 will be available to business users and developers through Google's Vertex AI and AI Studio.
The standard version of Gemini Pro will be updated to 1.5 Pro with a 128,000-token context window for general users.
Google is testing Gemini 1.5's safety and ethical boundaries, especially concerning its large context window.
3. OpenAI, in collaboration with Microsoft Threat Intelligence, halted malicious AI cyber-activities by state-affiliated groups from China, Iran, North Korea, and Russia.
4. Andrej Karpathy, a founding member of OpenAI, has left OpenAI to explore personal interests in AI and technology.
5. OpenAI enhances ChatGPT w/ new memory controls that allow users to manage what the AI remembers, enhancing privacy and while increasing personalization.
6. ElevenLabs launches Voice Actor Payouts, enabling users to monetize their authentic voice clones in the Voice Library, along w/ expanding their Speech-to-Speech feature to multiple languages.
7. Nvidia's 'Chat with RTX' brings local AI chatbot capabilities to PCs, offering private, instant data analysis without cloud reliance, while showcasing AI's potential in summarizing and searching personal documents and YouTube videos
8. Slack has launched Slack AI, a suite of built-in AI features that provide summaries of threads, channel recaps, and allow users to ask questions about work-related topics.
9. AI giants, (including Google, Microsoft, Meta, OpenAI, Adobe, and TikTok) form an accord to address the use of AI-generated content that could deceive voters in elections, outlining steps to mitigate risks, such as labeling suspected AI content and educating the public on AI dangers.
10. Court partially dismisses the copyright lawsuit against OpenAI by Sarah Silverman and others, but the direct infringement claim remains.
11. The White House is developing a method to cryptographically verify official communications from President Joe Biden to distinguish them from AI-generated deepfakes.
12. Cohere launched Aya, a revolutionary multilingual AI model and dataset, covering 101 languages with a massive open-source model, aiming to enhance AI's understanding and generation capabilities across a broad spectrum of languages.
13. Apple is internally testing new generative AI features for Xcode and Spotlight
14. Stability AI introduces Stable Cascade, a new text-to-image generative AI model, promising more efficient and flexible performance than Stable Diffusion.
15. Google launched its new AI hub in Paris, to accommodate around 300 researchers and engineers already working for Google Research, DeepMind, YouTube, and Chrome.
16. Protesters gathered outside OpenAI's office, organized by Pause AI and No AGI, to oppose the company's development of military AI and AGI, partly in response to OpenAI altering its usage policy to allow AI for military purposes and taking the Pentagon as a client.
17. Mozilla announces major downsizing and refocus on Firefox and AI.
18. Airbnb's acquires GamePlanner.AI in a strategic move to integrate AI into its platform, focusing on creating a new, adaptive AI interface to revolutionize user experience with the 'ultimate concierge’
💰 Top Funding News:
1. Lambda, which is building a robust GPU cloud infrastructure for AI, raised a $320M Series C led by US Innovative Technology Fund (USIT), w/ B Capital, SK Telecom, T. Rowe Price Associates, Inc., Crescent Cove, Mercato Partners, 1517 Fund, Bloomberg Beta, and Gradient Ventures, among others.
2. GitHub launched the next cohort of its GitHub Accelerator program, focusing on AI and aiming to support projects and products that leverage AI advancements under an open source license to improve the world.
3. LangChain, the open-source LLM app development framework, launched LangSmith, the first paid LLMOps product for general availability, and raised a $25M Series A led by Sequoia Capital.
4. Sierra, an AI agents startup co-founded by former Salesforce co-CEO Bret Taylor and ex-Google VP Clay Bavor, raised $110M in funding from Sequoia Capital and Benchmark.
5. Klaus, an Estonian startup founded in 2019 to support customer service agents, has been acquired by Zendesk, for an undisclosed amount after previously raising $19.3M.
6. Xensam, a provider of AI-powered Software Asset Management (SAM), raised $40M in growth funding from Expedition Growth Capital.
7. Marqo, which enables the creation of advanced AI-powered search experiences w/ their next gen vector search of unstructured enterprise data, raised a $12.5M Series A led by Lightspeed w/ Blackbird VC, January Capital, and others.
8. Together Computer Inc, which operates cloud platforms geared toward AI workloads and provides access to high-end NVIDIA GPUs for companies, is seeking more than $100M in new funding, w/ Salesforce Inc.’s VC arm expected to lead the round and Coatue expected to participate.
9. Upstash, a serverless data platform tailored for data-intensive applications w/vector databases specifically for generative AI applications, raised a $10M Series A led by Andreessen Horowitz (a16z).
10. Rasa, a startup offering infrastructure to enable developers at large enterprises to build robust generative conversational AI assistants, raised a $30M Series C co-led by StepStone Group and PayPal Ventures, w/ Andreessen Horowitz (a16z), Accel, and Basis Set Ventures.
11. Clarity, an AI cybersecurity startup aimed at fighting deepfakes, raised a $16M Seed Round led by Walden Catalyst and Bessemer Venture Partners.
13. Mindy, an email-based executive assistant startup aiming to simplify information retrieval through AI automation, raised $6M in a Seed Round led by Sequoia Capital and Founders Fund.