Friday, October 18, 2024

Techpresso logo

TECHPRESSO

Cracks appear in Microsoft and OpenAI partnership

Receive a daily summary of what happened in tech, powered by ML and AI.

Thank you! We sent you a verification email.
Oops! Something went wrong while submitting the form.
Join 1,500+ thinkers, builders and investors.

Good morning! This is your daily โ˜•๏ธ Techpresso.



In today's Techpresso:

๐Ÿ‘€ Cracks appear in Microsoft and OpenAI partnership

๐Ÿฝ๏ธ Meta fires staff who abused $25 meal vouchers

๐ŸŽง Google's AI podcast generator gets major updates

๐Ÿข AWS CEO tells workers to quit if they don't want to come back to the office

๐Ÿ”’ X updates privacy policy to allow third parties to train AI models

๐Ÿ’ต US Treasury uses AI to recover billions from fraud

๐ŸŽ + 11 other news you might like

๐Ÿ”ฎ + 3 handpicked research papers and tools

๐Ÿ‘€ Cracks appear in Microsoft and OpenAI partnershipLINK

  • OpenAI and Microsoft's once strong partnership is showing signs of strain as OpenAI seeks more computing power and reduced costs while Microsoft hedges its bets by investing in competing technologies.
  • Tensions have grown between the two companies, with disagreements over resource allocation, Microsoft hiring talent from OpenAI competitors, and protocol breaches by Microsoft engineers.
  • Despite recent adjustments to the partnership, such as OpenAI securing additional computing resources from Oracle, OpenAI is still facing significant financial pressures, expecting to lose $5 billion in 2024.
  • ๐Ÿฝ๏ธ Meta fires staff who abused $25 meal vouchersLINK

  • Meta fired around two dozen employees in Los Angeles for abusing $25 meal vouchers, using them to purchase non-food household items like toothpaste and tea, despite some having salaries as high as $400,000.
  • The staff had been misusing the system for a long time, sometimes pooling funds or ordering meals to their homes, even though the vouchers were meant for office use.
  • This action is part of Meta's broader restructuring efforts, which have included multiple rounds of layoffs affecting thousands of workers over the past two years.
  • ๐ŸŽง Google's AI podcast generator gets major updatesLINK

  • Google's NotebookLM now allows users to guide AI-generated audio conversations by focusing on specific topics, improving upon the previous automatic approach of generating holistic audio summaries.
  • Over 80,000 organizations have adopted NotebookLM, prompting Google to launch a business pilot program offering early access to new features, training, and support for participating companies.
  • NotebookLM saw a traffic increase to 4.17 million monthly visits, with Google planning to expand its functionality with features like customizable audio summaries, different voices and languages, and a potential native mobile app.
  • ๐Ÿข AWS CEO tells workers to quit if they don't want to come back to the officeLINK

  • Amazon CEO Matt Garman stated in an all-hands meeting that employees who are unhappy with the new five-day office return mandate should leave the company.
  • The policy, gradually increasing in-office attendance requirements to five days a week by January 2025, has met resistance from staff, who previously opposed a three-day office return plan.
  • Garman believes remote work hinders innovation and criticizes the hybrid model for not facilitating effective collaboration, emphasizing the importance of in-person team interactions.
  • ๐Ÿ”’ X updates privacy policy to allow third parties to train AI modelsLINK

  • X is updating its privacy policy to allow third-party collaborators to use users' data for training artificial intelligence models, effective November 15, 2024.
  • The policy mentions an opt-out feature for data sharing, but currently, there is no clear way for users to exercise this option, as noted by TechCrunch.
  • Besides privacy policy changes, X is imposing stricter penalties for scraping, charging $15,000 for accessing over one million posts a day, to protect user data and resources.
  • ๐Ÿ’ต US Treasury uses AI to recover billions from fraudLINK

  • The Treasury Department is using artificial intelligence to combat fraud, successfully preventing or recovering over $4 billion in potential losses within the last year.
  • Officials have emphasized the use of machine learning to analyze extensive data sets, effectively identifying patterns of fraud frequently utilized by criminals.
  • The department plans to extend its AI-developed resources and insights to other federal agencies to bolster their anti-fraud measures.
  • Other news you might like

    Google moves Gemini app team to AI research lab DeepMind.LINK

    Perplexity lets you search your internal enterprise files and the web.LINK

    Sam Altmanโ€™s Worldcoin becomes World and shows new iris-scanning Orb to prove your humanity.LINK

    US investigates 2.4m Tesla self-driving vehicles after reported collisions.LINK

    Microsoft said it lost weeks of security logs for its customersโ€™ cloud products.LINK

    Intel seeks billions for minority stake in Altera business, sources say.LINK

    Stripe in talks to acquire Bridge for $1 billion.LINK

    Bluesky gains half a million new users in one day after X changes how blocking works.LINK

    University student builds simple raycaster maze demo with transparency support in Microsoft Excel.LINK

    Man who threw away $500M Bitcoin hard drive sues city for right to search landfill.LINK

    23andMeโ€™s entire board resigned on the same day. Founder Anne Wojcicki still thinks the startup is savable.LINK

    Latest research and tools

    Grandmaster-Level Chess Without Search: trains a transformer model on 10 million chess games to play at grandmaster level, surpassing traditional engines and AlphaZero, without using explicit search algorithms or domain-specific adjustments.LINK

    Microsoft BitNet: an official inference framework designed to perform fast, energy-efficient, and lossless inference on 1-bit Large Language Models, notably improving performance and reducing energy usage on CPUs, with plans for NPU and GPU support.LINK

    LLMD: a Large Language Model for Interpreting Longitudinal Medical Records: the paper introduces a language model designed to analyze and interpret complex, long-term medical records.LINK


    Want to get the latest news differently? Find us on:

    twitter instagram instagram instagram


    See you tomorrow for a new dose of โ˜•๏ธ Techpresso!

    Feeling behind on AI?

    You're not alone. Techpresso is a daily tech newsletter that tracks the latest tech trends and tools you need to know. Join 150,000+ professionals from top companies like OpenAI, Apple, Google and more. 100% FREE.
    Thank you! We sent you a verification email.
    Oops! Something went wrong while submitting the form.
    Join 1,500+ thinkers, builders and investors.
    You're in! Thanks for subscribing to Techpresso :)
    Oops! Something went wrong while submitting the form.
    Join 5,000+ thinkers, builders and investors.