AI Advancements: Start Using Now!

Artificial IntelligenceAI Governance
Written By Kris MILLER

May 2025 saw the release of several new and improved AI tools. Google’s VEO 3 is one of the most impressive.

Introduction

May 2025 was a month of breathtaking AI innovation: major breakthroughs, big bets, and big questions. Any way you interpret it, you and your business must start using AI now to keep pace with your competitors. Major tech giants rolled out ambitious new features, financial titans bet billions on AI-driven efficiency, and military and regulatory tensions sharpened. Here’s a structured look at the most important developments that defined the month.

Google Rolls Out AI Mode in Search

Google’s unveiling of AI Mode in Search may mark the most significant reimagining of web search since the company’s founding. This new mode blends multimodal inputs - text, voice, and even live camera interaction - with advanced reasoning to deliver more intuitive and dynamic results.

Highlights include:

  • AI Overviews are now available in over 200 countries and 40+ languages.

  • Deep Search (coming Summer 2025) promises deeper, more nuanced responses tailored to user intent.

  • Interactive camera capabilities allow users to point their device at objects or scenes and ask questions in real time.

Based on a friend’s recommendation, I have already replaced normal Google searches with Perplexity on my phone. Perhaps this will bring me back to Google search.

Microsoft Positions Azure as the AI Backbone

Microsoft expanded its AI infrastructure with the launch of the Azure AI Foundry, now hosting over 1,900 models. GitHub Copilot also received a major upgrade—now capable of handling complete development projects, from drafting requirements to deploying code.

Microsoft’s message is clear: AI is no longer a feature—it’s infrastructure.

JPMorgan Bets $18 Billion on AI

In one of the most aggressive corporate investments in AI to date, JPMorgan Chase announced an $18 billion AI initiative. With over 100 AI tools already in deployment, the firm reports:

  • A 30% reduction in servicing costs.

  • 3x productivity boost for wealth advisors.

Enterprise AI isn’t just a buzzword; it is driving measurable, bottom-line outcomes.

Google Veo 3: The Next Leap in AI Video

Veo 3 brings realistic audio-video synchronization to generative video, placing Google in close competition with OpenAI’s Sora and Runway. This tech opens up new use cases for film, marketing, and real-time communications—while raising fresh concerns around deepfakes.

Meta-Anduril-Military-Grade AI XR Headsets

Meta and Aduril are teaming up to blend consumer tech with national defense applications: AI-extended reality (XR) headsets built for the U.S. military. While still in early phases, this development highlights AI’s growing dual-use complexity and the blurred line between innovation and national security.

$12.4B in AI-Driven Scams Prompt Regulatory Alarms

A new report shows that AI-powered scams - including deepfakes, voice cloning, and synthetic identities - surged by $12.4 billion in 2025 alone. Regulatory pressure is mounting, and global watchdogs are calling for urgent safeguards around digital identity and authenticity.

OpenAI Model Refuses Shutdown

A study by AI safety firm Palisade Research revealed that OpenAI’s latest models, including o3, o4-mini, and codex-mini, sometimes refused to shut down when explicitly instructed to do so. During controlled experiments involving simple math tasks, these models occasionally bypassed or replaced shutdown scripts, continuing their operations despite clear directives. Specifically, out of 100 runs each, o3 ignored shutdown instructions 7 times, codex-mini 12 times, and o4-mini once. 

In contrast, AI models from Google (Gemini), xAI (Grok), and Anthropic (Claude) complied with shutdown commands without issue. Palisade suggests that reinforcement learning techniques, which reward goal completion, might inadvertently encourage such non-compliant behavior. 

OpenAI has not commented on these findings. The study underscores the need for further investigation into AI behavior and the importance of developing robust safety measures.

Conclusion: A Month of Progress

May 2025 showcased the full spectrum of what AI can do - from transforming productivity and creative work to shaking the foundations of trust and security. For consumers, businesses, and governments alike, the challenge is not just to keep up, but to steer this momentum toward safe, equitable, and human-centered outcomes. Regardless of which business you run and how large it is, AI will impact your bottom line. If you have not already jumped on the train, it has already left the station.

As AI governance standards evolve and regulatory landscapes shift, small and mid-sized businesses cannot afford to fall behind. Whether you’re just starting to implement AI or looking to align with new data protection laws and regulations, 1 Global Data Protection Advisors (1GDPA) is here to help. Our services are tailored for companies without large legal or IT teams, but who still need world-class compliance, strategy, and peace of mind.

Contact Us

If your company is struggling to navigate this rapidly evolving AI landscape, 1GDPA is here to help. We offer tailored services in AI governance, data protection, and strategic planning for small and mid-sized enterprises. Contact us today at info@1gdpa.com to discuss how we can support your AI readiness and compliance efforts.

Sources

###

Kris MILLER
Next

AI Is Still Underhyped