Each week, we bring you 5 stories that resonated the most in our internal Slack channel #AI-news. We write the newsletter using various AI tools because we're an AI company and our marketing wants to move with the times too. π
Today you're reading the 64th issue in a row.
#1
Meta introduces Llama 4 models, surpassing Google and OpenAI in benchmarks π§ π!
Meta has unveiled Llama 4, claiming its models outperform Google and OpenAI in key benchmarks π. Llama 4 Scout, which fits on a single Nvidia H100 GPU, outperforms Google Gemini 3 and Mistral 3.1 β‘π». Llama 4 Maverick competes with GPT-4 and Gemini 2.0 Flash, utilizing fewer parameters (individual elements used by the model for learning and predictions) π§©π§.
The upcoming Llama 4 Behemoth, with 288 billion active parameters, promises to surpass GPT-4.5 in STEM tasks (science, technology, engineering, and math) π¬.
The new "mixture of experts" architecture enhances efficiency by activating only the necessary parts of the model for specific tasks βοΈ. While Meta labels the model as open-source, its license requires permission for commercial entities with over 700 million users, raising questions about its true open-source nature π. More details will be revealed at LlamaCon on April 29 π π.
Β
#2
$40 billion investment from Microsoft and SoftBank boosts OpenAIβs valuation to $300 billion π»π‘!
OpenAI has secured $40 billion in funding, marking the largest investment ever received by a private tech company π°π.Β
This investment values the company at $300 billion, placing it among the top private companies in the world, just behind SpaceX and TikTokβs parent company, ByteDance π.The funding came from SoftBank ($30 billion) and a group of investors, including Microsoft π»π‘.
OpenAI plans to use these funds for advances in AI research and to scale its infrastructure π§ . The deal also includes a condition that OpenAI must transition to a for-profit organization by December 31, which could impact its future funding π΅π. With 500 million weekly ChatGPT users, OpenAI is poised for further expansion π.
Β
#3
Trump's new tariffs could disrupt AI chip production and pricing π₯οΈπ§.
New tariffs from President Trump have caused a drop in the stock prices of semiconductor companies like Nvidia, TSMC, and AMD ππ». Starting April 9, the U.S. will impose a 34% tariff on imports from China and 32% on imports from Taiwan π¨π³πΉπΌ, which could raise the price of AI chips, potentially impacting the demand for GPUs used in AI servers π₯οΈβ‘.
However, some analysts believe that companies working on Artificial General Intelligence (AGI) will continue purchasing these chips, even if prices rise π€. TSMC, which manufactures some of the worldβs most advanced semiconductors, may be particularly affected, as it supplies major U.S. companies like Nvidia and Apple ππ».
While itβs unclear whether the tariffs will bring more chip production to the U.S., itβs expected that importing chips from Taiwan and China will become more expensive for companies πΈπ.
Β
#4
Microsoft celebrates 50th anniversary and introduces new features for AI assistant Copilot ππ€.
Microsoft is celebrating its 50th anniversary by rolling out new features for Copilot, its AI assistant ππ€. The assistant now remembers personal details, such as birthdays, helps with presentations π, and can even book services π .
Itβs designed to simplify and streamline life by being proactive π‘. This move keeps Microsoft competitive in the AI space, where companies like OpenAI and xAI are leading βοΈπ§ .
Copilot also now personalizes recommendations, such as podcasts π§ and shopping suggestions π, adding value for everyday users π©βπ»π‘. With these updates, Microsoft is not just creating a tool but improving AI intelligence and making it more helpful in daily life ππͺ. What do you think π€?
Β
#5
New tariff calculation resembles AI recommendations, raising concerns about AI-driven policy π§ π‘!
Trump's new trade policy uses a controversial tariff calculation that closely aligns with the formula recommended by AI chatbots, such as ChatGPT π§ π‘. This formula involves dividing the countryβs trade deficit by its exports to the U.S., and then halving the result π. Economists have criticized this approach for being overly simplistic and potentially harmful to global trade ππΌ.
Chatbots like ChatGPT, Gemini, Grok, and Claude all suggested similar formulas to balance the trade deficit, with some even recommending halving the tariff to achieve a "reasonable" result π‘π. While it's unclear whether the White House used AI for this decision, the striking similarity between their formula and chatbot suggestions raises concerns about AI-driven policy π€ποΈ.
Although the chatbots warned that real-world trade dynamics are far more complex, that didnβt stop the U.S. from proceeding with its plan πΊπΈπ. Now that the tariffs have taken effect, the world is watching to see how they will impact the global market and trade relations ππ. What do you think π€?
Β
