PLUS: Gemini gets a major personalization upgrade
Good morning, AI enthusiasts. OpenAI just revealed its vision for AI governance — seeking copyright exemptions for themselves while pushing to restrict Chinese open-source rivals like DeepSeek.
As the $500B Stargate architect flexes its growing influence in Washington, has OpenAI's political strategy become as calculated as its AGI roadmap?
-
OpenAI pushes for federal shield in AI Action Plan
-
Cohere’s new efficient enterprise AI model
-
Turn your health data into personalized insights
-
Gemini taps into Google history with personalization
-
4 new AI tools & 4 job opportunities
OPENAI

Image source: Getty Images
The Rundown: OpenAI just published its 15-page submission for the White House’s request for public input towards the AI Action Plan, seeking to shield AI companies from state regulations in exchange for federal oversight.
-
OpenAI warns that the 781 state-level AI bills introduced this year could hinder American innovation and competitiveness against China's AI ambitions.
-
The proposal includes additional calls for infrastructure investment, copyright reform, and expanding access to government datasets for AI development.
-
They notably called out China’s “unfettered access to data”, calling the race for AI ‘effectively over’ if fair use copyright laws are not applied in the U.S.
-
OpenAI also pushed for the U.S government to ban models like DeepSeek due to security risks, calling the lab “state-controlled”.
Why it matters: Between its $500B Stargate Project and growing influence in Washington, OpenAI's regulatory ambitions now rival its technical ones. But the push for state law exemptions while advocating to restrict open-source rivals comes at an awkward time given heavy criticism over closed-source models and copyright battles.
TOGETHER WITH TURING
The Rundown: Attending GTC next week or Bay Area based? Turing and Fireworks AI invite you to join the Generative AI Gathering in San Jose on Wednesday, March 19, from 5-7 PM PT.
-
Global networking with leaders from the AI/LLM communities
-
Discussions ranging from AI-assisted software development to the future of LLM training
-
An evening of creative cocktails and fabulous food with your peers
COHERE

The Rundown: Cohere just unveiled Command A, a new enterprise-focused AI model that matches top competitors' performance while running on just two GPUs — also featuring strong multilingual capabilities and a large context window.
-
Command A achieves 156 tokens per second, running 1.75x faster than GPT-4o and 2.4x faster than DeepSeek-V3 while requiring just two GPUs.
-
Cohere’s model also matches or surpasses GPT-4o and DeepSeek-V3 across human evaluations for business, STEM, coding, and agentic tasks.
-
The model features a 256k context window, support for 23 languages, and specialized enterprise features like advanced RAG capabilities.
-
Command A will also integrate into Cohere’s North platform, allowing enterprises to securely deploy agents with their own internal databases.
Why it matters: Much of the AI space is chasing higher benchmarks, but Cohere's efficiency push could be ripe for enterprise adoption. The ability to run competitive, agentic AI capabilities on minimal hardware not only cuts costs but also makes private deployments more practical for companies concerned about data security.
AI TRAINING

The Rundown: In this tutorial, you’ll learn how to use Perplexity Spaces to create a personalized AI health assistant by uploading your health documents and setting custom instructions.
-
Visit Perplexity, click on "Spaces" in the left bar, and select "Create a Space."
-
Name your health space, add a description, and set custom AI instructions to guide how the AI interprets your questions and what context it prioritizes when generating responses.
-
Upload your health documents including medical records, medication lists, and wellness tracking data (do not include sensitive data you don’t want to share).
-
Start asking health questions that leverage your personal data for truly customized insights.
Note: While your personalized health assistant can provide valuable insights, it should never replace professional medical advice. Consult healthcare providers if any significant findings emerge from your analysis.
PRESENTED BY INNOVATING WITH AI
The Rundown: Innovating with AI's new program, AI Consultancy Project, transforms AI enthusiasts into professional consultants — tapping into a market projected to reach $54.7B by 2032.
The 6-month program delivers:
-
Proven frameworks for client acquisition and service delivery
-
A step-by-step path to six-figure consulting income
-
Students who land their first AI client in as little as 3 days

The Rundown: Google just released new personalization features for its Gemini AI assistant, allowing the AI to tap into users' Search history and eventually other Google apps to deliver more tailored responses and contextually aware conversations.
-
The experimental feature uses the Gemini 2.0 Flash Thinking model to analyze when personal data could enhance responses.
-
Google is starting with user’s search history, with plans to expand to apps like Google Photos and YouTube for additional data insights.
-
Users maintain control through opt-in permissions and the ability to disconnect their history at any time, with the feature restricted to users over 18.
-
Free users can also now access Gems (custom chatbots) and improved Deep Research capabilities previously limited to Advanced subscribers.
Why it matters: Google is leveraging its massive web of user data to create a more personalized experience — which also comes with the tricky dance of leveraging personal info without violating user trust. But with prominent opt-ins and a user base already used to connecting via apps, Google is well-positioned to pull it off.
-
🤖 Gemma 3 - Google’s multimodal, multilingual, 128k context AI model family
-
⚡️ Gemini 2.0 Flash exp - Upload, create, and edit images directly via text conversations with new native image generation
-
👁️ R1-Omni - Alibaba’s new open-source multimodal reasoning model that can ‘read’ emotions
-
🧠 Perplexity MCP - Connect Perplexity’s Sonar model to Claude for real-time web search capabilities
AI2 released OLMo 2 32B, the first fully open model to outperform GPT-3.5 and GPT-4o mini on academic benchmarks while requiring just a third of the training compute used by comparable models like Qwen 2.5 32B.
Microsoft and Xbox introduced "Copilot for Gaming," an AI-powered assistant designed to help players get into games faster, provide in-game coaching, and enhance social experiences — with early access coming soon to mobile users.
Alibaba launched New Quark, a redesigned version of the company’s AI assistant mobile app now powered by its new Qwen reasoning model.
Google added new support for YouTube in the Gemini API and Google AI Studio, allowing the model to use vision capabilities to interact with video content.
Former OpenAI researcher Andrej Karpathy predicted that AI will usher in a fundamental shift in web content design, with pages being optimized for LLMs and context windows over human readability.
Insilico Medicine secured $110M funding for its AI-powered drug discovery platform, which includes multimodal foundation models and a fully automated robotic lab with a bipedal humanoid AI Scientist.
Join our next workshop today at 4 PM EST to learn about how to use Manus AI to boost productivity and automate everyday tasks easily with Dr. Alvaro Cintas, The Rundown’s AI professor.
We’ll always keep this newsletter 100% free. To support our work, consider sharing The Rundown with your friends, and we’ll send you more free goodies.
Rowan, Joey, Zach, Alvaro, and Jason—The Rundown’s editorial team
[ad_2]
Source link