Keeping an Eye on AI – March 2025 Highlights

March came in hot with waves of innovation, tech drama, bold launches, and some big questions about the future of artificial intelligence. Every week, something shifts. I’ve been tracking it all—here’s your complete rundown. Keeping an Eye on AI

OpenAI Delays Voice Cloning (Again)

One year after announcing it, OpenAI has once again postponed the launch of its voice cloning tool. No new date was given, and the delay seems linked to ethical concerns about misuse. The debate on responsible AI rages on.

AI Agents: The Trend That Took Over the Month

AI agents were the talk of the town. OpenAI plans to charge up to $20,000/month for specialized agents and released an API to create highly personalized agents. Experts say this market is about to explode—and I agree.

Microsoft Launches Dragon Copilot for Healthcare

Dragon Copilot is a clinical AI assistant that listens to doctor-patient conversations and turns them into standardized medical documentation. Microsoft enters the healthcare space with a bang.

Copilot for Gaming: AI Comes to Xbox

Xbox introduced a voice-activated AI assistant to enhance the gaming experience. For now, it’s only available in the Xbox mobile app, but it opens new doors for gamers.

OpenAI Requests Ban on DeepSeek in the U.S.

OpenAI formally asked U.S. authorities to consider banning DeepSeek, a Chinese AI model, citing security and IP risks. The global geopolitical race in AI is heating up.

DeepSeek Rejects VC Investment

DeepSeek refused outside capital to retain full control of its business. It continues to be funded by the High-Flyer fund, managed by its founder. Confidence or strategy? Maybe both.

Google’s Gemini 2.5 is on the Way

The Gemini 2.5 model will introduce “pause to think” capabilities, mimicking more human reasoning. No launch date yet, but early access is available to platform subscribers.

Google Adds AI Mode to Search

Google introduced an AI Mode in its Search, enabling more nuanced, real-time answers. Currently, it’s exclusive to Google One AI Premium subscribers.

Amazon Developing Hybrid Reasoning AI

Amazon confirmed it’s working on an AI model based on hybrid reasoning, combining symbolic logic with machine learning. It’s expected to launch in June.

Mistral Launches Multimodal PDF API

Mistral released an API that extracts text from PDFs while also identifying images and illustrations. A game-changer for document-heavy workflows.

Microsoft Secures Renewable Energy for AI Expansion

To keep up with growing AI demand, Microsoft signed a deal for an additional 475 megawatts of solar energy. Three new solar projects will power its data centers through AES.

Meta Tests AI Comments on Instagram

With the feature "Write with Meta AI," Instagram now suggests comments based on posted images. In a time when we crave authenticity, will AI-generated comments really work?

Meta’s $800M Bid for FuriosaAI Rejected

Meta offered $800 million to acquire South Korean chipmaker FuriosaAI, but the deal was rejected. The startup is currently seeking $48 million in funding independently.

n8n Raises $60M for AI Automation Expansion

Workflow automation platform n8n raised $60 million to grow in markets like the U.S. It’s quickly becoming a major player in intelligent automation.

Manus AI’s Browser Use Feature Goes Viral

Manus AI's tool allows AI agents to autonomously navigate the web. It’s revolutionizing web accessibility and automation.

Neo Gamma: Humanoid Robot Tests Begin

Norwegian company 1X announced real-world testing of its Neo Gamma humanoid robot in residential homes this year. A key step in human-AI coexistence through data collection.

ChatGPT Image Generation Gets a Boost

GPT-4o now lets users create and edit images right in ChatGPT. Everyone’s jumped on the trend—and yes, I did too 😄.