OpenAI and Microsoft are investigating claims that DeepSeek illegally trained its AI model using data from OpenAI's technology without authorization, potentially violating OpenAI's terms of service.
 
Wednesday 29 January 2025 | Join Free | Upgrade

Together with Sponsor

Hi there, this is your daily ☕️ Techpresso.

In today's Techpresso:

💥 OpenAI says DeepSeek used its models illegally

✈️ Boom's XB-1 becomes first civil aircraft to go supersonic

🛰️ Apple and SpaceX bring Starlink support to iPhones

🛡️ OpenAI launches ChatGPT for government agencies

💳 X strikes deal with Visa in push toward 'everything app' goal

🔓 Apple chips can be hacked to leak secrets from Gmail, iCloud, and more

🎁 + 8 other news you might like

🔮 + 6 handpicked research papers and tools

FROM OUR PARTNER


⛈️ Is your cloud spend out of control? 💸 Learn how FinOps can help you cut costs without sacrificing innovation.
Sponsor

Ever wondered how companies keep cloud spend under control as AI and other high-demand workloads send costs through the roof?

The secret is with FinOps-a framework that brings engineering, finance, and operations together to maximize cloud efficiency and reduce your cloud bill.

This guide breaks down the core principles of FinOps, offering expert insights and a practical checklist to help you get started.

Inside, you'll learn:

🤑 What FinOps is and why it matters.

🏗 Key strategies to structure your FinOps practice.

✅ A practical checklist for implementing FinOps principles in your organization.

📖 Download the Guide for Free 📖

💥 OpenAI says DeepSeek used its models illegally LINK

  • OpenAI and Microsoft are investigating claims that DeepSeek illegally trained its AI model using data from OpenAI's technology without authorization, potentially violating OpenAI's terms of service.
  • The technique of "distillation," where one AI learns from another, is central to the accusation, with OpenAI alleging that DeepSeek has extracted knowledge from its models to create a competitive language model.
  • OpenAI is facing criticism for its complaint against DeepSeek, as it mirrors its own practices of using large amounts of data from various sources, which it defends as fair use under copyright law.

✈️ Boom's XB-1 becomes first civil aircraft to go supersonic LINK

  • Boom Supersonic's XB-1 demonstrator aircraft became the first civil plane to break the sound barrier in a test flight without government funding or direction, achieving Mach 1.122.
  • The aircraft's successful 35-minute flight took place at California's Mojave Air & Space Port, where it surpassed Mach 1 multiple times before landing safely back at the base.
  • XB-1's milestone flight paves the way for Boom's larger Overture airliner, planned for commercial service by the end of the decade, with orders already from major airlines like United and American Airlines.

🛰️ Apple and SpaceX bring Starlink support to iPhones LINK

  • Apple has collaborated with SpaceX and T-Mobile to integrate Starlink satellite texting capabilities into iPhones via the new iOS 18.3 update, as reported by Bloomberg.
  • T-Mobile is conducting a limited beta test for iPhone users in the United States to try out the Starlink service, with plans to expand device compatibility in the future.
  • Currently, the Starlink service supports texting only, but SpaceX and T-Mobile plan to add data and voice features later, while Apple's satellite service operates in several countries through Globalstar.

🛡️ OpenAI launches ChatGPT for government agencies LINK

  • OpenAI has introduced "ChatGPT Gov," a tailored version of its AI assistant, specifically designed for U.S. government agencies to enhance their operations with advanced AI capabilities.
  • The initiative allows federal, state, and local authorities to implement ChatGPT Gov through Microsoft's Azure Cloud or Azure Government Cloud, ensuring compliance with security and privacy standards.
  • Over 90,000 users across 3,500 government agencies, including the Air Force Research Laboratory and Los Alamos National Laboratory, are already utilizing OpenAI's services, highlighting its growing governmental influence.

💳 X strikes deal with Visa in push toward 'everything app' goal LINK

  • X has partnered with Visa to introduce 'X Money,' a financial service enabling direct payments and digital wallet integration, furthering Elon Musk's vision of transforming X into a multifunctional app.
  • Visa Direct will allow users to instantly fund their digital wallets and link debit cards, making peer-to-peer transactions and bank transfers straightforward and efficient on the platform.
  • Launching in late 2025, X Money will compete with services like Venmo and Zelle, aiming to integrate diverse functionalities similar to WeChat within a single app environment.

🔓 Apple chips can be hacked to leak secrets from Gmail, iCloud, and more LINK

  • Apple chips used in Macs, iPhones, and iPads have two new security flaws that expose private data such as credit card details and locations from Chrome and Safari browsers on specific websites.
  • The vulnerabilities target the A- and M-series chip sets and involve side channel attacks, exploiting the speculative execution process that predicts CPU behavior to enhance performance.
  • The FLOP and SLAP attacks manipulate different speculative execution features, allowing attackers to access sensitive information like email contents and location history by predicting incorrect memory values and addresses.

Other news & articles you might like

  • Boost efficiency with Fyxer AI: Save 60 minutes daily by automating emails and meetings. Prioritize key messages, send personalized drafts, and get meeting summaries.Set up in 30 seconds!
  • DeepSeek Shows Meta's A.I. Strategy Is Working LINK
  • Jack Dorsey's Block has an AI agent too LINK
  • France's answer to ChatGPT fails spectacularly, goes offline after three days of bizarre responses LINK
  • Hugging Face wants to reverse-engineer DeepSeek's R1 reasoning model LINK
  • Figure AI details plan to improve humanoid robot safety in the workplace LINK
  • Garmin users say their watches are bricked with a 'blue triangle of death' LINK
  • LinkedIn removes accounts of AI 'co-workers' looking for jobs LINK
  • AI haters build tarpits to trap and trick AI scrapers that ignore robots.txt LINK

Latest research and tools

Proof of Concept WMI Virus: a zero-day exploit that uses Windows Management Instrumentation to store and self-extract malicious code without writing to the disk, allowing privilege escalation and evading antivirus detection. LINK

Transformer Lab: an open-source app that lets users experiment with large language models, build and use datasets, and run locally or in the cloud through a cross-platform GUI. LINK

Auto-Differentiating Any LLM Workflow: this paper discusses a new method that enables language models to create optimized workflows without the need for manual prompting. LINK

3D Scene Reconstruction in Adverse Weather Conditions via Gaussian Splatting: the paper discusses a new method that improves 3D scene reconstruction in difficult weather conditions using a technique called Gaussian splatting. LINK

Self-Replicating AI: researchers are experimenting with AI systems that can autonomously replicate themselves in controlled laboratory environments. LINK

Algebraic Databases: the paper discusses how algebraic structures can be applied to database theory to improve data processing and querying. LINK

Want to get the latest news differently?

Not subscribed to ☕️ Techpresso yet? 

☕️ Techpresso

Have a feedback? Send me a message.

You were sent this message because you subscribed to Techpresso.

Unsubscribe