Genie 3 Pushes AI Agents From Text Into Real-Time Simulated Worlds

In partnership with

The Latest Agentic Development

Google rolls out Project Genie to paying subscribers powered by Genie 3

Google DeepMind launched Project Genie, an experimental web prototype powered by the Genie 3 world model that lets Google AI Ultra subscribers in the U.S. create, explore, and remix interactive 3D worlds from text or image prompts in real time. Unlike static content generation, Project Genie generates environments as the user moves through them, demonstrating Google’s first consumer-facing use of its world-model tech and signaling how agent training/simulations could soon shift to dynamic environments.

Security wake-up call: AI agents seen as “untrustworthy without identity”

Keyfactor released new research showing that 86% of cybersecurity professionals say AI agents and autonomous systems cannot be trusted without unique, dynamic digital identities, and 85% expect such identities to be as common as human or machine identities within five years. The report highlights an “Agentic AI Security Recognition-Action Gap”: while most security leaders recognize agent-specific risks, only about half have agent governance frameworks in place and just 28% believe they could actually stop a rogue agent once deployed. Strategically, this positions identity, PKI, and policy-as-code infrastructure as mandatory layers in any scaled agent platform, and elevates agent identity to a board-level concern for enterprises deploying tools that trigger payments, modify data, or orchestrate other agents.

Moltbot’s viral surge becomes an early “agent security” stress test

Axios reports rapid adoption of the local-first open-source agent Moltbot (formerly Clawdbot), which runs with broad machine permissions (shell/files/browser/email/calendar). A security researcher found hundreds of exposed/misconfigured control panels online (risking leaked API keys/credentials and remote command execution), and Token Security said 22% of its customers already have employees using Moltbot underscoring why “agent ops” security is quickly becoming mandatory, not optional.

Rezolve AI pitches “Agentic Checkout” as protocol-aligned transaction infrastructure

Rezolve AI claims strategic alignment with emerging agentic commerce protocols and positions its Agentic Checkout as a standards-based, auditable layer for AI-initiated purchases. It cites forecasts projecting $1T–$5T in annual agentic commerce transaction value by 2030, arguing that shared standards + delegated payment models + identity controls will be the unlock for large-scale agentic transactions.

Agentic commerce goes mainstream: BCG says “agents in charge”

BCG X published a Jan 29 analysis arguing that 2026 will be the year of the personal agent, as all major consumer AI platforms roll out standards for how businesses expose offers, inventory, and pricing to AI assistants. The piece describes “agentic commerce” as a world where AI shopping agents research options, compare prices, apply preferences and constraints, assemble baskets, and increasingly complete purchases on behalf of consumers—shifting commerce from channels to embedded “surfaces” and from human-operated flows to agent-authored briefs. Strategically, BCG recommends brands decide whether to play a “destination” game (own channels) or an “evaluation” game (being chosen by agents), emphasizing that product data quality, pricing, availability, and fulfillment terms will determine whether agents surface or skip an offer.

What to Watch this Week

 🎙️AI on Fire: Real Builders. Real Heat.

Stories from people building AI Agents. Explore all episodes here

🏆 The Leaderboard Never Sleeps

The global ranking of AI agents is shifting every day. Who’s on top? Who just dropped?

👉 Explore the Leaderboard →

🗺️ The Map of AI Agents (Live & Growing)

We’re charting the entire AI agent ecosystem — thousands of options across categories.
Your next agent is already on the map.

👉 Explore the Landscape Map →