Daily AI Brief (Feb 18, 2026)

Copilot data bug, Meta’s Nvidia chip megadeal, and Gemini’s music push.

TL;DR

• Microsoft confirmed a Copilot Chat bug involving exposure of confidential Office email summaries, intensifying enterprise focus on data controls.
• Meta signed a multiyear Nvidia infrastructure deal spanning millions of AI chips, reinforcing compute as a strategic moat.
• Google added Lyria 3 music generation to Gemini, signaling continued expansion into multimodal creation.
• World Labs raised major funding with Autodesk backing to push world models into practical 3D workflows.
• OpenAI expanded higher-ed partnerships in India, highlighting AI talent-pipeline competition.
• Perplexity reportedly shifted away from ad focus toward paid products, underscoring trust-linked monetization choices.

1) Microsoft confirms Copilot email exposure bug

Microsoft confirmed a bug that allowed Copilot Chat to summarize confidential Office email content incorrectly, including cases where protections should have limited processing.
Source: TechCrunch

Takeaway: Enterprise buyers will likely demand stronger proof of data controls and clearer separation between productivity data and model ingestion paths.

2) Meta commits to a major Nvidia chip expansion

Meta signed a multiyear infrastructure deal with Nvidia spanning millions of AI chips across current and next-generation CPU/GPU lines.
Source: The Verge

Takeaway: AI infrastructure spending remains structurally high, and compute access is still a strategic moat.

3) Google adds Lyria 3 music generation to Gemini

Google is rolling out AI music generation in Gemini, allowing users to create short tracks from prompts and media references.
Source: The Verge

Takeaway: Leading assistants are evolving into multimodal creation suites, not just chatbot interfaces.

4) World Labs secures strategic backing for 3D world models

World Labs raised significant funding with Autodesk participation to push world-model capabilities into practical 3D and design workflows.
Source: TechCrunch

Takeaway: Spatial AI is increasingly positioned as an enterprise productivity layer, especially in design-heavy industries.

5) OpenAI expands education footprint in India

OpenAI announced partnerships with major Indian higher-education institutions to broaden AI skills and integration.
Source: TechCrunch

Takeaway: The race for AI adoption now includes academia and workforce pipelines at national scale.

6) Perplexity reportedly moves away from ads

Perplexity executives reportedly said the company is stepping back from ad focus to prioritize paid products.
Source: The Verge

Takeaway: AI product trust is becoming directly tied to monetization choices.

Bottom line

1. Trust and governance are becoming product-defining.
2. Compute concentration remains a core strategic lever.
3. Real-world integration is where the next adoption wave is being won.