Nadella: “Treat AI Agents Like Employees”

Nadella: Give AI Agents Identities, Sandboxes, and Policies — “Treat Them Like Employees”

Microsoft CEO Satya Nadella dropped the clearest playbook yet for managing AI agents at scale. Speaking on Reid Hoffman’s “Possible Podcast” on Friday, he laid out three non-negotiable requirements: identities, sandboxes, and policies. “You need to give them identities, you need to give them sandboxes, then you need to set policies to govern them,” Nadella said — framing agents not as tools but as a digital workforce that needs HR-style management.

Why it matters for your business: If the CEO of Microsoft — who personally manages 100 AI coding agents — says the cognitive load is “extremely high,” that’s a signal. The companies that win won’t just deploy agents; they’ll build governance systems around them. Microsoft’s Agent 365 suite (including Entra for digital identity and Purview for data labeling) is the infrastructure play. For SMBs, the takeaway is clear: before you scale agents, define who they are, what they can touch, and how you’ll audit them.

Tencent Cloud Launches Enterprise Agent Toolkit Across 20 Verticals

Tencent Cloud unveiled its first systematic “Efficiency Agent Toolset” in Beijing this week, offering personalized AI agents for finance, healthcare, manufacturing, and 17 other sectors. The suite includes WorkBuddy and Enterprise CodeBuddy, signaling a strategic shift from model supremacy to real-world agent deployment. For businesses watching the global AI race: China’s tech giants are moving fast from “we have the best model” to “here’s an agent that does your job.”

Source → TechnoSports

Small Business Owners Are Now Managing Teams of AI Agents — Not Just Using Tools

Entrepreneur reports a quiet shift: SMB owners are moving from “using AI as a tool” to overseeing entire AI agent teams. A bankruptcy lawyer profiled uses agents for client intake, document prep, and routine correspondence. His honest assessment: “AI will, probably sooner than later, be able to do what I do, for a lot cheaper.” The workforce management model Nadella described at enterprise scale is already trickling down to Main Street.

Source → Entrepreneur

Anthropic Drops Bombshell: Calls for Global AI Slowdown, Warns of “Recursive Self-Improvement”

In a blog post Thursday, Anthropic warned that rapid advances could enable AI systems to improve themselves without human intervention — what it calls “recursive self-improvement” — far sooner than expected. The company proposed a global agreement to slow or temporarily pause frontier AI development, arguing society needs time for “alignment research to keep up.” The timing is notable: Anthropic recently restricted its own powerful “Mythos” cybersecurity model. Critics suggest the warnings are self-serving, but the core concern is landing with policymakers already spooked by bioweapons risk.

Source → Hindustan Times

Google Inks $920M/Month SpaceX Compute Deal — “Demand Meaningfully Exceeding Supply”

In a regulatory filing dated June 5, Google revealed it will pay SpaceX $920 million per month from October 2026 through June 2029 for access to roughly 110,000 Nvidia GPUs. CEO Sundar Pichai told investors that AI product demand is “meaningfully exceeding our available supply.” The deal — worth roughly $30 billion total — comes just weeks after Google’s cloud pact with Anthropic and days before SpaceX’s IPO. The compute crunch is real, and Big Tech is paying unprecedented sums to feed it.

Source → The New York Times

Altman, Amodei, and Suleyman Unite: Joint Letter to Congress Warns AI Is Lowering Bioweapons Barriers

In an extraordinary show of unity, the CEOs of OpenAI, Anthropic, and Microsoft AI co-signed a letter to Congress warning that AI is “meaningfully eroding” the knowledge barriers that prevent bad actors from designing biological weapons. Organized by the Foundation for American Innovation, the letter urges mandatory screening for synthetic DNA/RNA orders — backing the Biosecurity Modernization and Innovation Act of 2026. When rivals set aside competition to jointly ask for regulation, it’s worth paying attention.

Source → Fortune