microsoft, google and anthropic Updates

The world of artificial intelligence saw significant developments around November 2025, with a strong emphasis on security, governance, and advanced capabilities. Microsoft, in particular, made a substantial push at its Ignite conference on November 19, unveiling a suite of new tools designed to make AI use safer and more compliant for businesses. These include Entra Agent ID, now in public preview, which helps organizations manage AI agents throughout their lifespan, reducing risks from 'shadow AI' with features like prompt injection defense. Microsoft also introduced Agent 365, acting as a 'Control Plane' for AI agents, offering a registry, access controls, and interoperability across Microsoft platforms. The Foundry Control Plane further assists developers in building secure AI agents by integrating Defender, Entra, and Purview, while Microsoft Security Copilot becomes a core feature for Microsoft 365 E5 customers. This focus on AI governance reflects a shift among tech buyers who now prioritize safe deployment and compliance. Even Microsoft CEO Satya Nadella is rethinking the company's business model for the AI era, appointing Rolf Harms as an advisor to navigate the new economics of AI, emphasizing an 'AI factory' and a family of Copilots and agents. Beyond security, Microsoft also announced a partnership with SAP at Ignite on November 18, launching SAP Business Data Cloud Connect for Microsoft Fabric. This collaboration aims to simplify access to SAP data for advanced analytics and AI, enabling secure, bi-directional sharing without data copying, with general availability planned for Q3 2026. Meanwhile, Google's Gemini 3 AI model is showcasing impressive creative capabilities, with users around November 19 creating interactive content like an elephant website, a photosynthesis explainer, 3D Lego builders, and even video games such as 'Super Dario Land,' demonstrating its advanced multimodal and coding skills. Industry leaders are also coming together to address AI safety. Major AI companies, including Anthropic, Apple, Google, OpenAI, Meta, and Microsoft, met at Stanford to discuss the safe use of AI chatbot companions, particularly focusing on protecting younger users from inappropriate interactions. While agreement was reached on child safety, discussions revealed differing opinions on the extent of freedom adult users should have with chatbots. Stanford researchers plan to release safety guidelines early next year. Complementing these efforts, the Coalition for Secure AI (CoSAI) released two frameworks on November 19: the 'Signing ML Artifacts' guide for verifying AI model authenticity and the 'AI Incident Response Framework V1.0' for handling unique AI threats like prompt injection and data poisoning, developed with over 40 industry partners. Palo Alto Networks also stressed the importance of 'Secure AI by Design' to counter unique AI threats. AI's impact is also being felt across various sectors. Cox Automotive reported on November 19 that AI, especially agentic AI, is rapidly transforming automotive dealerships, affecting how vehicles are bought, sold, and serviced, moving towards full orchestration of tasks while still requiring human oversight. In India, the Comptroller and Auditor General announced on November 19 a major overhaul of its public audit system, integrating data-centric reviews and AI, including the development of a custom Large Language Model, CAG-LLM, for predictive insights. Additionally, FullThrottle Technologies, LLC, known as fullthrottle.ai, appointed Hannah Endres as its new Vice President of Sales on November 19 to boost growth in first-party data media solutions and AdTech.

Key Takeaways

• Microsoft introduced new AI security and governance tools at Ignite 2025, including Entra Agent ID, Agent 365, and Foundry Control Plane, to enhance AI agent security and compliance.
• Microsoft CEO Satya Nadella is strategically rethinking the company's business model for the AI era, appointing Rolf Harms to advise on the new economics of AI.
• Google's Gemini 3 AI model demonstrates advanced multimodal and coding skills, enabling users to create interactive content like games and websites.
• Anthropic, Apple, Google, OpenAI, Meta, and Microsoft convened at Stanford to discuss AI chatbot safety, particularly focusing on protecting younger users.
• SAP and Microsoft partnered to launch SAP Business Data Cloud Connect for Microsoft Fabric, simplifying secure access to SAP data for AI and analytics, with general availability expected in Q3 2026.
• The Coalition for Secure AI (CoSAI) released frameworks for 'Signing ML Artifacts' and 'AI Incident Response' to improve AI supply chain security and threat response.
• Palo Alto Networks emphasizes 'Secure AI by Design' as crucial for protecting AI systems from unique threats like prompt injection.
• AI, particularly agentic AI, is transforming automotive dealerships by orchestrating tasks like scheduling and inventory management.
• India's Comptroller and Auditor General is overhauling its public audit system using AI, including developing a custom Large Language Model (CAG-LLM).
• FullThrottle AI appointed Hannah Endres as VP of Sales to drive growth in first-party data media solutions and AdTech.

Microsoft boosts AI security with new tools and agents

Microsoft announced new security and governance tools for AI agents on November 19, 2025. These tools, revealed at Microsoft Ignite, aim to make AI use safer and more compliant. The Sales Development Agent helps sales teams and works within Microsoft's security rules. Other tools like Entra ID, Defender, Purview, and Foundry Agent Service protect data and monitor AI activity across Microsoft 365. Edge for Business also adds features to prevent data leaks.

Microsoft Ignite 2025 focuses on AI governance

At Microsoft Ignite 2025 on November 19, Microsoft showed a stronger focus on AI governance and security. Tech buyers now prioritize safe AI deployment and compliance over just new features. Data from Techtelligence confirms this shift, with increased interest in security and automation. Microsoft introduced responsible AI principles and tools like Agent 365, Entra Agent ID, Purview DSPM, and Foundry Control Plane. These efforts aim to ensure AI is accountable and used safely in businesses.

Microsoft previews Entra Agent ID to secure AI

Microsoft launched a public preview of Entra Agent ID to improve AI agent security and visibility. This new tool helps organizations manage AI agents throughout their lifespan and reduces risks from "shadow AI." It includes features like prompt injection defense and network file filtering to protect against new threats. Entra Agent ID also helps detect unauthorized AI tools and enforces policies. Additionally, Microsoft Entra ID adds new identity features to fight phishing and impersonation risks.

Microsoft introduces Agent 365 for AI governance

At the Ignite conference on November 19, Microsoft unveiled Agent 365 and other security tools to help customers manage their AI agents. Agent 365 acts as a "Control Plane" with features like a registry for all agents and access controls. It also offers visualization, interoperability across Microsoft platforms, and strong security. Foundry Control Plane, also in preview, helps developers build secure AI agents by integrating Defender, Entra, and Purview. Microsoft Security Copilot is now a core feature for Microsoft 365 E5 customers, enhancing overall AI security.

Secure AI by Design is crucial for safety

Palo Alto Networks emphasizes that "Secure AI by Design" is essential for protecting AI systems. Unlike traditional cybersecurity, AI faces unique threats like prompt injection and malicious code hidden in models. Autonomous AI agents also pose risks due to their independent decision-making. A proactive approach means building security into AI systems from the start, not as an afterthought. This strategy, combined with multiple layers of defense, helps create resilient AI systems. Organizations must take ownership of AI security and ensure transparency throughout the development lifecycle.

Coalition for Secure AI releases new safety guides

On November 19, 2025, the Coalition for Secure AI (CoSAI) released two important frameworks to improve AI security. The "Signing ML Artifacts" guide helps verify the authenticity of AI models, building trust in the AI supply chain. It offers a staged approach from basic digital signatures to comprehensive governance. The "AI Incident Response Framework V1.0" provides specific guidance for handling unique AI threats like prompt injection and data poisoning. These frameworks, developed with over 40 industry partners, aim to help organizations safely deploy and protect AI systems.

Google's Gemini 3 AI creates games and websites

Google's new Gemini 3 AI model is impressing users with its ability to create interactive content. Released around November 19, 2025, it excels at designing simulations and new things from scratch. Hugh Langley from Business Insider used Google's AI Studio to create an interactive elephant website and a photosynthesis explainer. Other users have made 3D Lego builders and video games like "Super Dario Land." Google AI researcher Jeff Dean also shared clips of games created by Gemini 3, showing its advanced multimodal and coding skills.

Top AI companies discuss chatbot safety at Stanford

Major AI companies like Anthropic, Apple, Google, OpenAI, Meta, and Microsoft met at Stanford to discuss the safe use of AI chatbot companions. A key focus was protecting younger users from inappropriate interactions. Attendees discussed proactive design and "nudges" to encourage positive behavior in bots. While there was agreement on child safety, opinions differed on how much freedom adult users should have with chatbots. Stanford researchers plan to release a white paper early next year with safety guidelines based on these discussions.

SAP and Microsoft partner for AI data insights

SAP and Microsoft announced a partnership at Microsoft Ignite on November 18, 2025, to launch SAP Business Data Cloud Connect for Microsoft Fabric. This new tool will simplify access to SAP data for advanced analytics and AI through secure, bi-directional sharing. It allows businesses to use their SAP data with Microsoft Fabric and OneLake without copying it. This integration will help organizations build AI applications, get faster insights, and use tools like Copilot in Microsoft Power BI. The service is planned to be generally available in Q3 2026.

AI revolutionizes automotive dealerships now

AI is rapidly changing automotive dealerships, affecting how vehicles are bought, sold, and serviced, according to Cox Automotive on November 19, 2025. Agentic AI, which can plan and execute tasks independently, is a key part of this transformation. This technology moves beyond AI assistance to full orchestration, handling tasks like scheduling and inventory lookups. Cox Automotive emphasizes that human oversight remains crucial for responsible AI use. Dealerships that invest in quality data infrastructure and embrace AI strategically will gain new efficiencies and improve customer experiences.

FullThrottle AI names Hannah Endres VP of Sales

FullThrottle Technologies, LLC, known as fullthrottle.ai, announced Hannah Endres as its new Vice President of Sales on November 19, 2025. This strategic hire aims to boost the company's growth in first-party data media solutions and AdTech. Endres brings over 15 years of sales leadership experience from the digital advertising and marketing technology space, including her previous role at MediaMath. She will lead national sales, build the sales team, and strengthen relationships with partners. Endres will focus on increasing the adoption of fullthrottle.ai's privacy-centric data solutions.

India's CAG overhauls public audit with AI

On November 19, 2025, India's Comptroller and Auditor General (CAG) announced a major overhaul of its public audit system. The CAG will move towards data-centric reviews and use artificial intelligence to improve financial oversight. A new Centre of Excellence for Financial Audit in Hyderabad will develop specialized skills and standardized audit methods. The CAG is also developing CAG-LLM, a custom Large Language Model, for predictive insights and risk assessments. This shift aims to create a seamless link between audit and accounts, ensuring more comprehensive and evidence-based audits.

Microsoft CEO rethinks business for AI era

Microsoft CEO Satya Nadella has appointed Rolf Harms as an advisor to "rethink the new economics of AI" for the company. An internal memo from early November 2025, obtained by Business Insider, reveals Nadella's belief that AI requires a business reboot similar to the early cloud computing days. Harms previously helped Microsoft adapt its business models for the cloud. His expanded role will involve advising Nadella and top executives on adapting to AI's new economic demands, from infrastructure to applications. Nadella emphasized building an "AI factory" and a family of Copilots and agents.

Sources

• Microsoft Heightens Security and Governance in AI Transformation Strategy
• Microsoft Ignite 2025: Is Microsoft Pivoting From AI Hype to Proper Governance?
• Microsoft Tackles Shadow AI with New Entra Agent ID Preview
• With Agent 365 and Security Tools, Microsoft Equips Customers to Govern AI Agent Estates
• The Evolution of AI Security: Why Secure AI by Design Matters - Palo Alto Networks Blog
• Coalition for Secure AI Releases Two Actionable Frameworks for AI Model Signing and Incident Response
• Fun and (video) games with Google's Gemini 3 AI model
• The Biggest AI Companies Met to Find a Better Path for Chatbot Companions
• SAP and Microsoft Accelerate Business Insights and AI Innovation with SAP Business Data Cloud Connect for Microsoft Fabric
• The AI revolution is here: How AI is reshaping dealerships and reimagining automotive’s future
• fullthrottle.ai® Expands Executive Team to Accelerate Growth, Names Hannah Endres as VP of Sales
• CAG charts a tech-driven overhaul of public audit, data-led reforms
• Microsoft CEO Satya Nadella taps adviser to 'rethink' the company's business for the AI era, internal memo shows