AGI

The U.S. Department of Defense is expanding its trusted AI partners, adding Microsoft, Amazon, Nvidia, and Reflection AI. These firms, joining OpenAI, xAI, and Google, will have their AI technologies approved for classified military engagements under “any lawful use.” This broad clause previously led to a canceled contract with Anthropic AI over concerns about domestic surveillance and autonomous weapons. The Pentagon aims to prevent vendor lock-in and equip service members with advanced AI tools for strategic deployment.

Google is internally testing “Remy,” a new AI personal agent for Gemini. Remy aims to move beyond conversational responses to proactive task execution, acting autonomously on user behalf across various Google and third-party services. This development signifies Google’s strategic investment in agent technology, seeking to create more sophisticated and anticipatory AI assistants. User control and privacy are emphasized through features like the Gemini Privacy Hub.

Physical AI integrates AI into real-world systems, posing complex governance challenges. With industrial robot adoption soaring, the market for Physical AI is projected to expand significantly. Unlike software AI, physical systems interact directly with dynamic environments and human users, demanding stringent safety parameters and clear escalation protocols. Google DeepMind’s Gemini Robotics, built on embodied AI principles, exemplifies this shift, offering advanced capabilities for robot control and reasoning. Effective Physical AI requires generality, interactivity, and dexterity, alongside robust visual perception, spatial reasoning, and task planning. Safety controls, traditionally software-based, must now be embedded into system design for physical interactions, with frameworks like NIST and ISO adapting to this evolving landscape.

Google’s Gemini Enterprise Agent Platform, launched at Cloud Next ’26, integrates agentic AI governance as a core feature, addressing a significant industry gap. Unlike previous add-ons, this platform assigns cryptographic identities to agents and centralizes oversight through an Agent Gateway. This move tackles the widespread challenge of AI sprawl and lack of control, where many organizations struggle to manage AI deployments effectively. Google is shifting focus from model capabilities to control plane ownership, offering a robust solution for enterprises to build, scale, and govern AI agents securely.

Enterprise AI is shifting from aspirational to imperative, demanding near-perfect accuracy and robust governance. The move from 90% to 100% accuracy is existential, transforming LLMs into autonomous agents requiring rigorous management. Key challenges include agent sprawl, data foundation readiness, and intent-based interfaces. True enterprise intelligence must leverage proprietary data and structured relational models, not just generic LLMs. Competitive defense emerges from customer-specific AI, requiring embedded functionality, agentic orchestration, and industry-specific intelligence, all underpinned by strong governance.

GitHub is updating Copilot with an “AI Credits” system, moving from per-query pricing to a value-based model. This change utilizes “tokens” to measure AI processing, with both input prompts and generated code consuming them. While pricing tiers remain, users will receive AI Credits instead of query limits. One credit is valued at one cent, with Copilot Pro offering 1,000 credits monthly. Token costs vary based on LLM, query complexity, and model cache. Core features like code completions will remain free.

LG and NVIDIA are reportedly in preliminary discussions exploring collaborations in physical AI, data center solutions, and mobility. LG aims to leverage NVIDIA’s processing power for its advanced hardware, particularly in thermal management for AI data centers and low-latency inference for home robots. NVIDIA could gain access to LG’s mass-market data and distribution channels for AI model training, while both companies could benefit from unifying automotive infotainment with NVIDIA’s autonomous driving platforms.

API gateways are crucial for managing enterprise data security and governance with AI. They act as a central control point for authentication, logging, and access control, enabling organizations to track AI tool data requests and permissions. However, gateways are network-layer defenses, like firewalls, and cannot inherently prevent software-layer vulnerabilities from LLMs or code. A holistic security strategy encompassing application and AI model integrity is essential beyond perimeter defenses to mitigate risks.

Australian financial regulators are flagging significant deficiencies in AI governance at financial firms. A recent review found boards are often overly reliant on vendor information and lack a deep understanding of AI risks, such as unpredictable model behavior and operational impact. APRA stresses the need for clearer AI strategies aligned with risk appetite, robust monitoring, error remediation, human oversight in high-risk decisions, and stronger cybersecurity measures. Dependencies on single AI providers are also a concern.

EMEA enterprises face AI rollout challenges, with many projects stalled due to execution issues and the need for financial validation. Boards are re-evaluating AI investments, demanding concrete ROI beyond traditional metrics. Success hinges on aligning AI with human workflows, robust infrastructure, strong governance, and a commercial mindset from technology leaders to drive tangible business outcomes and revenue growth.