OpenAI has a new secret project
Plus more whistleblowers against OpenAI
Today’s Highlights:
📰 News: Open has a new project, code-name ‘Strawberry’ + Whistleblowers file a complaint w/ SEC against OpenAI
💰 Funding: Fireworks AI and Regard both raise $60M+ Series B’s
⚡️ Top News Stories:
1. OpenAI is developing a new AI technology under the code name "Strawberry," aimed at enhancing reasoning capabilities to enable advanced autonomous tasks like deep research and long-horizon tasks.
The workings of Strawberry are tightly guarded within OpenAI, but it's designed to improve AI's ability to plan ahead and navigate complex problems more effectively.
Strawberry aims to enable AI to conduct autonomous web research and perform tasks typically handled by software and machine learning engineers by using post-training methods similar to Stanford's Self-Taught Reasoner.
2. Whistleblowers have filed a complaint with the SEC, alleging OpenAI illegally barred employees from reporting risks of its technology by enforcing restrictive non-disclosure agreements.
OpenAI's employment contracts allegedly required staff to waive federal whistleblower compensation rights and seek company approval before disclosing information to federal authorities, violating federal laws.
OpenAI, in a statement, claimed its policies protect employees’ rights to make disclosures and emphasized its commitment to rigorous debate and recent policy changes to address concerns.
The letter urges the SEC to investigate and enforce penalties, demanding OpenAI notify employees of their rights and ensure their agreements comply with federal laws to promote transparency and safety in AI development.
The complaint highlights broader issues with tech companies' use of NDAs to silence employees, particularly in AI, where federal regulation is still developing.
3. Google's Gemini AI is being used to significantly enhance the capabilities of its RT-2 robots, improving their navigation and task completion through better understanding and interaction.
Gemini-powered robots achieved a 90% success rate in completing over 50 user instructions in a large, 9,000-square-foot area, demonstrating significant improvement in task performance.
Here’s a video of the demo:
4. Amazon's AI-powered shopping assistant, Rufus, is now available to all U.S. customers through the Amazon Shopping app, where it helps users make informed decisions by answering questions about product details, customer reviews, and recommendations based on user needs.
5. Lattice, an AI-powered HR software, attempted to treat AI bots as employees by giving them official records and work-related responsibilities, such as onboarding and goal setting, which was met by immediate backlash as industry professionals expressed confusion and criticism, questioning the practicality and reasoning behind integrating AI as employees in human resources processes.
6. Microsoft unveiled SpreadsheetLLM, an AI system designed to work with spreadsheets, enhancing enterprise productivity by understanding and manipulating spreadsheet data.
7. Google introduces Vids, a productivity app in Workspace Labs, enabling users to create videos as easily as making slides by allowing users to combine docs, slides, voiceovers, and videos into a timeline for presentation creation. Here is video of the demo:
💰 Top Funding News:
1.Standard Bots, which integrates advanced AI into robotics systems, enabling robots to learn tasks through demonstration and adapt to various environments, raised a $63M Series B, led by General Catalyst, w/ Amazon Industrial Innovation Fund and Samsung Next.
2. Regard, which utilizes AI to analyze patients' health data, aiding physicians in diagnosing conditions more accurately and efficiently by mining thousands of data points from medical charts, raised a $61M Series B, led by Oak HC/FT, w/ Cedars-Sinai Health Ventures, TenOneTen, Calibrate Ventures, and Techstars.
3. Fireworks AI, which develops compound AI systems that enable developers to create multi-model and multimodal AI applications quickly and efficiently, raised $52M Series B, led by Sequoia Capital, w/ NVIDIA, AMD, MongoDB Ventures, Benchmark, Databricks Ventures, and several prominent tech executives.
4, Kindo AI, which focuses on securing enterprise AI by providing centralized management, visibility, and compliance for AI models, raised a $20.6M Series A, led by Drive Capital, w/ RRE Ventures, Marlinspike Partners, Riot Ventures, Eniac Ventures, New Era Ventures, and Sunset Ventures.