Microsoft AI chief Mustafa Suleyman says there are three labs that matter Microsoft announced significant AI advancements at its Build conference, including proprietary reasoning models and a super app, signaling a strategic shift to develop its own AI capabilities independently of OpenAI. The company aims to rival top AI labs by building advanced, multimodal frontier models from scratch, emphasizing enterprise security and compliance. Microsoft also showcased a new cybersecurity tool and is integrating AI agents like OpenClaw into its ecosystem, leveraging its existing client base and resources to compete in the evolving AI market.

• Microsoft is launching new AI initiatives, including a super app, in-house reasoning models, and a cybersecurity tool.
• The company aims to become one of the top four AI labs globally, building its own frontier models from scratch.
• Microsoft unveiled MAI-Thinking-1, its first reasoning model, built independently and emphasizing performance and cost-effectiveness.
• A new AI cybersecurity tool, MDASH, uses 100 AI agents to identify exploitable bugs.
• Microsoft is committed to supporting and integrating AI agents like OpenClaw into its Windows ecosystem.
• The company is developing a Copilot ‘super app’ that integrates various AI assistants, including ‘Autopilots’ for enterprise users.
• Microsoft emphasizes ‘humanist superintelligence’ and provides security tools and guardrails for its AI offerings.
• Nvidia’s RTX Spark chip is supporting Microsoft’s AI agent development.
• Microsoft aims to leverage its existing client base, reputation for security, and financial resources in the AI race. Continue reading https://www.theverge.com/ai-artificial-intelligence/942242/microsoft-build-ai-agents-openai-competition