OpenAI Eyeing $90B as Valuation

Author

Ts Yotov
September 27, 2023

In today’s edition:

🗞️ News

📰 Cloudflare launches new AI tools to help customers deploy and run models

📰 OpenAI Eyeing $90B as Valuation

💡 Small bits - Microsoft Researchers Propose AI Morality Test for LLMs and more

📚 Nerd section - GPUs from all major suppliers are susceptible to a new "pixel-stealing" attack and more

Cloudflare is introducing a new set of AI tools designed to help customers develop, deploy, and run AI models efficiently at the network edge.

  • Workers AI offers customers access to nearby GPUs for AI model operations on a pay-as-you-go model.

  • Vectorize serves as a database for storing vector embeddings, crucial components for machine learning algorithms.

  • AI Gateway gives insights into AI traffic, tracking metrics like model request duration, user count, and overall AI app costs. It also offers cost-saving features like caching and rate limiting.

  • Cloudflare’s AI solutions aim to make AI more accessible to developers and businesses, leveraging their existing infrastructure for improved performance and reduced costs.

Why it Matters: Cloudflare’s foray into AI tools simplifies AI integration for developers, emphasizing cost savings and efficiency. This move reflects the growing importance of AI across various business applications and Cloudflare's commitment to streamline its adoption.

OpenAI is considering selling shares, potentially elevating its valuation from $29 billion to $80-$90 billion.

  • The potential share sale allows employees to sell their current shares instead of the company issuing new ones.

  • Earlier this year, OpenAI secured over $300 million in funding at a $29 billion valuation.

  • Microsoft made a significant investment in OpenAI, roughly estimated at $10 billion.

  • ChatGPT, OpenAI's generative AI assistant, has been a major tech success, with upcoming features for voice interactions.

  • OpenAI, 49% owned by Microsoft, anticipated $1 billion in 2023 revenue.

Why it Matters: OpenAI's rapidly growing valuation signifies the immense potential and success of AI technologies, notably ChatGPT, in the modern tech landscape.

💡 Small Bits 💡

🕵️ CIA are developing an AI chatbot for better intelligence access, rivaling China's AI advancements.

🎁 Google.org is donating $10 million to FIRST and the REC Foundation to help 300,000 U.S. middle school students engage in robotics programs, blending traditional skills with emerging tech like AI.

🤖 Former Tinder CEO Renate Nyborg's startup Meeno, targeting loneliness with an AI "personal mentor" chatbot, has secured $3.9 million in seed funding from Sequoia and aims to launch in various countries by December.

🧍‍♂️ An AI-generated image inaccurately depicting the iconic "Tank Man" from Tiananmen Square taking a selfie became the top result on Google Search, raising concerns over the trustworthiness of search results and the potential spread of misinformation.

⚙️ Jony Ive is reportedly developing an AI gadget with OpenAI’s Sam Altman: Ive and Altman might be designing a mysterious AI hardware device, with previous collaborations hinting at their shared vision for the future of AI tech.

👁️‍🗨️ OpenAI’s GPT-4 with vision still has flaws, paper reveals: Despite its touted multimodality, GPT-4's image-analyzing capabilities can be problematic, misinterpreting images, occasionally showing biases, and even misunderstanding harmful symbols or creating inappropriate content.

🪶 The writers strike is over; here’s how AI negotiations shook out: Writers Guild of America finalizes a deal with studios which limits AI use in writing, ensuring writers aren't displaced by technology and their works aren't exploited for AI training.

🧭 Microsoft Researchers Propose AI Morality Test for LLMs: Using Kohlberg's model and the Defining Issues Test, they found most large language models exhibit intermediate moral reasoning, with only GPT-4 nearing advanced moral stages.

📚 Nerd section 📚

🐛 GPUs from all major suppliers are susceptible to a new "pixel-stealing" attack, where malicious websites can extract visual data, like usernames an, from other sites. The vulnerability, named GPU.zip, exploits GPU data compression techniques for performance, allowing attackers to bypass critical web security protocols.

Thank you for reading today’s edition!

We love to hear back from you!

Feel free to reply to our emails with questions, suggestions, or topics you'd like to see covered, or drop us a message on Twitter or Facebook.

Until tomorrow,
- Tsvetelin (Bits and Neurons)Amazon Brings Generative AI To Alexa And Fire TV