Welcome back to Daily Zaps, your regularly-scheduled dose of AI news ⚡️ 

Here’s what we got for ya today:

• 🇺🇸 $500 billion in private sector AI infrastructure

• ❌ Trump cancels Biden executive order on AI safety

• 🇨🇳 Chinese “reasoning” model rivals OpenAI o1

• 🤖Trump Admin Accused of Using AI to Draft Executive Orders

Let’s get right into it!

Sign Up | Advertise | Tools

GOVERNMENT

$500 billion in private sector AI infrastructure

President Trump is expected to announce a major private sector investment in U.S. artificial intelligence infrastructure, with OpenAI, SoftBank, and Oracle collaborating on a joint venture called Stargate. The initiative, starting with a data center project in Texas, plans to invest $100 billion initially, with up to $500 billion projected over the next four years.

Key executives, including SoftBank CEO Masayoshi Son, OpenAI's Sam Altman, and Oracle's Larry Ellison, are set to join the announcement at the White House. Additional investors may join the venture, but their identities have not yet been disclosed. This ambitious project aims to strengthen AI infrastructure across the United States.

GOVERNMENT

Trump cancels Biden executive order on AI safety

President Trump has rescinded President Biden's 2023 executive order on artificial intelligence, removing government constraints on AI development to foster faster innovation. The move, praised by industry leaders and Trump supporters, contrasts with Biden’s focus on AI oversight, which included testing requirements and safety evaluations to address potential risks. Critics of Biden’s order argued it hindered innovation with excessive regulation, while supporters highlighted the need for safety measures.

Trump's more permissive approach comes amid significant AI advancements in China, with firms like DeepSeek claiming breakthroughs rivaling U.S. tech. Conservative groups, such as Americans for Prosperity, applauded the decision as a step toward empowering private-sector leadership and maintaining U.S. global competitiveness. However, Trump’s administration will need to address Biden’s parting rule on AI export controls, which takes effect in May 2025.

WORLD NEWS

Chinese “reasoning” model rivals OpenAI o1

Chinese AI lab DeepSeek has released its new R1 model family, including the largest version with 671 billion parameters, under an open MIT license. The R1 models, designed for simulated reasoning (SR), aim to replicate human-like chains of thought during problem-solving and have shown strong performance on math and coding benchmarks, reportedly surpassing OpenAI’s o1 model in several areas. The release also includes smaller "DeepSeek-R1-Distill" models, ranging from 1.5 billion to 70 billion parameters, which can run on local hardware, making advanced reasoning capabilities more accessible.

The AI community has taken note of this breakthrough, as open-weights models often lag behind proprietary ones like OpenAI’s o1, and this marks a shift in the availability of high-performing public models. However, the cloud-hosted version of R1 includes censorship aligned with Chinese regulations, while locally hosted versions remain free of such constraints.

GOVERNMENT

Trump Admin Accused of Using AI to Draft Executive Orders

Hours after his inauguration, President Trump signed numerous executive orders that drew criticism for their sloppy formatting, grammatical errors, and oddly formulaic language, sparking speculation that AI tools may have been used in their drafting. Legal experts noted issues such as repeated numbering in a list within an order about Alaska's natural resources and simplistic, "grade school-style" descriptions in another renaming the Gulf of Mexico as the Gulf of America.

While some argue this reflects poor human execution rather than AI involvement, the errors raise concerns about potential confusion and enforcement challenges. The administration has not confirmed whether AI was used, leaving experts to question whether the lack of quality stems from rushed processes or generative AI tools.