This week’s AI news spans big strategic moves, evolving regulations, and a growing need for emotional intelligence – from Microsoft’s shifting OpenAI strategy to new tools reshaping healthcare, legal work, and creative industries. We’re also seeing fresh urgency inside agencies and an FBI warning on deepfake audio scams. Here are the top stories marketers and business leaders should have on their radar.
Inside Satya Nadella’s Bet on AI – and Microsoft’s Race to Stay on Top
If you follow AI news, you know that Microsoft's fingerprints are all over the place, yet they're more known for under-the-hood and infrastructure things than high-performing models like OpenAI or Perplexity. In this feature, Microsoft CEO Satya Nadella talks about his strategic tightrope: pushing Copilot to the center of Microsoft’s enterprise and consumer platforms while reducing dependence on OpenAI. When Chinese startup DeepSeek released a high-performing, low-cost open-source model, Nadella responded not with resistance, but with integration – adding it to Azure’s growing portfolio. His long game isn’t about owning the smartest model – it’s about commoditizing AI and owning the infrastructure layer. With 1,900+ models available through Azure and internal models like MAI-2 in development, Microsoft is betting that scale, flexibility, and trust will win the next wave of AI competition.
🔗 Read more
Multi-Agent Protocol Aims to Become AI’s Universal Language
If you've used a tool like Zapier or Maker, you know the immense potential of connecting disparate systems so they can accomplish more together than they can alone. That idea, increasingly called "interoperability," is at the center of a new standard called the Multi-Agent Communication Protocol (MCP), and it could become a huge piece in facilitating enterprise AI. Backed by Microsoft, Intel, and Cerebras, MCP is designed to let AI agents from different vendors collaborate seamlessly. Instead of siloed systems, organizations could deploy cross-functional AI agents that share goals, communicate in natural language, and work together on complex tasks. It’s a major step toward interoperability – especially valuable for large enterprises juggling multiple models and platforms. For marketers and business leaders, it’s another signal that the future of AI will be less about tools – and more about orchestration.
🔗 Read more
Google’s Gemini Live Rolls Out Conversational AI Experience
A couple weeks ago, I drafted a blog post while driving using ChatGPT's Voice mode. It was a cool experience and further glimpse into what kinds of interactions "multimodal" AI will facilitate. Now, Google has gotten into the game with the launch of Gemini Live, a real-time, voice-enabled conversational interface for its Gemini AI model. The new feature enables natural back-and-forth dialogue, even letting users interrupt the model mid-sentence – making it feel more like a human conversation than a voice command system. This rollout is part of Google’s broader plan to embed Gemini deeply into Android, Search, and productivity tools. For anyone building digital experiences, this raises the bar: conversational AI isn’t just about speed – it’s about presence, tone, and trust. Google is betting that more natural interaction will be the next frontier in user engagement.
🔗 Read more
OpenAI Launches HealthBench to Measure AI in Medicine
Below all of the constant noise about OpenAI is the fact that they are actively and seriously investing in real products and services related to the delivery of healthcare and support of basic research. This was further proven last week when they introduced HealthBench, a new benchmark designed to evaluate how well large language models perform across clinical tasks like diagnosis, documentation, and treatment recommendation. Built on expert-reviewed datasets, HealthBench doesn’t aim for direct clinical deployment, but rather transparency – helping developers, hospitals, and regulators identify where models are ready, and where they’re not. For healthcare marketers and digital strategists, it’s a sign that we’re entering a new phase: not just using AI, but validating it. Trust and performance benchmarks like this will be essential as AI tools move closer to frontline care. Check out the back-and-forth I had with other healthcare marketers on LinkedIn about Healthbench.
🔗 Read more
If we've learned anything in the first months of the new Trump administration, it's that they are taking AI investment seriously, and are putting the focus squarely on open innovation in AI, as opposed to the more cautious approach taken by the Biden administration. So, no surprise last week when a new Republican-led bill proposed a 10-year moratorium on state-level AI regulation, shifting all AI oversight to the federal government. The move is meant to eliminate compliance complexity and support U.S. competitiveness against China. But critics argue it undercuts states’ ability to address real risks – like algorithmic bias, misinformation, or consumer privacy. With hundreds of AI bills introduced at the state level last year, this proposal could reshape the governance landscape overnight. For business leaders, it’s a reminder that AI isn’t just a tech issue – it’s becoming a deeply political one, with real regulatory consequences.
🔗 Read more
FBI Warns of Deepfake Audio Scams Targeting Americans
The FBI is warning businesses and consumers about a surge in AI-generated audio scams, where attackers use deepfake voices to impersonate government officials or company executives. The goal: trick targets into transferring money, credentials, or confidential information. As audio cloning becomes more accessible, even seasoned professionals can be fooled. For marketers and CX leaders, this is more than a security issue – it’s a brand trust issue. Organizations will need to double down on verification protocols and public education as the line between real and fake becomes harder to spot.
🔗 Read more
Is This the End of Agencies as We Know Them?
As I shared in my wrap-up of the Mirren Live conference I spoke at a couple of weeks ago in NYC, agencies are in a period of rapid transition, facing deep uncertainty, and quickly reshaping themselves to try to stay ahead of the AI wave. Now, Tom Head, co-founder of Lab13 and former agency CEO, issued a bold warning on LinkedIn: the agency world has “18 months – maybe” to reinvent itself around AI. Clients are already seeing results from AI tools that deliver speed, personalization, and scale—raising questions about why they should pay premium prices for manual agency work. Head says many agencies are still in denial, resisting the level of operational and strategic change AI demands. For agency leaders, the post is a gut check: evolve into an AI-native firm – or risk becoming obsolete. Tom had a great appearance recently on the Creative Agency Account Manager Podcast, where he talked about the agency of the future. Great stuff; check it out.
🔗 Read more
MIT Disavows Study Claiming Big Productivity Gains from ChatGPT
MIT has publicly distanced itself from a high-profile paper by one of its PhD students claiming ChatGPT significantly boosts productivity. The university cited methodological flaws and emphasized that the research was neither peer-reviewed nor endorsed by MIT faculty. The backlash reflects broader concerns about how AI “wins” are communicated – especially when early-stage findings are amplified by tech media or vendors. For business leaders looking to integrate AI, the message is to be cautious of sweeping claims and demand rigor. AI can be powerful – but not everything that glitters is peer-reviewed gold.
🔗 Read more