On July 28, 2026, Italian aerospace and defence group Leonardo announced a sweeping digital transformation program that will bring Microsoft 365 Copilot to roughly 50,000 employees. Dubbed “Digital Horizon – A New Era of Collaboration with Microsoft 365,” the initiative isn’t just another software rollout—it’s a carefully architected blueprint for scaling AI across one of Europe’s most security-conscious enterprises.
Inside the Digital Horizon rollout
The program covers the full Microsoft 365 collaboration stack: Teams, OneDrive, Outlook, and the Copilot AI assistant. According to Leonardo’s press release, the goal is to make work “more connected, integrated, collaborative, and secure,” replacing siloed legacy tools with real-time cloud-based workflows.
But the real headline isn’t the technology list—it’s how the company is weaving everything together. A “When to Use What” smart compass will guide employees through the tangle of chats, channels, email, and document repositories. Short training modules, advanced prompting education, and an introduction to AI agents all come baked into the project. A network of internal “Digital Pioneers” will act as peer mentors, circulating best practices across business units. As Leonardo’s Chief People & Organization Officer Antonio Liotti put it, “The new tools and Artificial Intelligence foster collaboration, information sharing, and innovation, reducing the time spent on repetitive tasks and allowing people to focus on those with greater added value.”
Why this matters beyond the defence sector
Leonardo isn’t a typical office-drone corporation. It builds helicopters, cyber systems, satellites, and advanced electronics for global customers. Document permissions, intellectual property boundaries, and export controls are not abstract policy discussions—they are daily operational realities. So when a company like this deploys an AI that can surface forgotten files, summarize sensitive email threads, and draft technical documents, the margin for error is razor-thin.
That’s precisely why IT teams everywhere should steal from this playbook. The risks that keep defence executives awake at night (oversharing, misclassification, AI-assisted accidental disclosure) are the same ones that threaten banks, hospitals, law firms, and any organization handling personal data. Digital Horizon treats AI not as a magic wand but as a capability that must be paired with people, process, and security discipline.
Training: the backbone of AI adoption
Technology rollouts routinely fail because employees are left to figure things out on their own. Leonardo’s program attacks that head-on. The training components include:
- Short, digestible modules rather than long workshops
- A “When to Use What” compass to eliminate tool confusion
- Prompting education—not to turn everyone into a prompt engineer, but to teach specificity, context, and when to verify outputs manually
- Piloting of AI agents with clear boundaries
The Digital Pioneers network is especially smart. It mirrors Microsoft’s own champion model, where early adopters become the go-to people for questions like “How do I write a project update without accidentally sharing classified details?” or “Is it safe to use Copilot to analyse this supplier contract?” That peer-to-peer learning scales far better than a top-down IT mandate.
Security: the make-or-break factor
Microsoft 365 Copilot inherits your existing permissions—it doesn’t invent new ones. That’s both a feature and a ticking time bomb. If your SharePoint sites are littered with “Everyone except external users” links, or if ex-employees still have access to Teams channels, Copilot will happily surface that content. Microsoft’s own guidance warns that oversharing is the number-one risk, because the AI “works within the permissions and access controls that already exist.”
For IT leaders, the lesson is clear: a Copilot rollout must start with a data hygiene sprint.
High-priority tasks include:
- Scanning for broadly shared links and fixing or removing them
- Applying sensitivity labels across key document libraries
- Reviewing membership of high-risk groups and teams
- Setting external sharing defaults to “specific people only” where needed
- Turning on audit logging and data loss prevention (DLP) policies
Leonardo’s emphasis on “intelligence and security” signals that this cleanup is already underway. For organisations following suit, Microsoft now provides a Copilot security dashboard that flags oversharing and compliance risks—but it only works if you’ve done the groundwork.
How we got here
Microsoft launched Copilot for Microsoft 365 in preview in early 2023, with general availability arriving in November 2023. Since then, enterprise uptake has surged, but many large rollouts have been tentative. Defence contractors have been especially cautious—until now. Leonardo’s move follows a broader trend of “AI plus security” rather than “AI versus security.”
Digital Horizon also reflects a maturation in how organisations think about generative AI. Instead of treating Copilot as a standalone chatbot, it’s being embedded into a deliberate workplace transformation that includes cloud migration, collaboration redesign, and cultural change. That’s a far cry from the “turn it on and hope for the best” approach seen in some early adopter waves.
What IT leaders should steal from this plan
If you’re considering a similar deployment, here’s a blunt checklist distilled from Leonardo’s framework and Microsoft’s own deployment playbooks:
- Audit your data before you enable AI – Identify and remediate overshared content. Use SharePoint access review tools and Microsoft’s Copilot security dashboard.
- Design a training arc, not a single webinar – Include basic tool literacy, prompting skills, and a clear “when to use what” decision tree.
- Stand up a champion network – Recruit volunteers from different departments, give them extra support, and let them answer peer questions.
- Start with a pilot group – Pick a handful of teams, measure outcomes (not just prompts), and refine your approach before scaling.
- Define what “good” looks like – Move beyond license counts. Track adoption habits, time saved on specific workflows, and—crucially—security incident data.
- Govern AI agents aggressively – As Copilot extends into task-specific agents, set rules about what they can access, who can create them, and how outputs get reviewed.
Outlook
Leonardo’s Digital Horizon programme is still in its early days, but the structure it has put in place will be closely watched. Expect other defence contractors and regulated industries to follow a similar template: tie Copilot to a broader cloud migration, invest heavily in training and peer champions, and treat security readiness as a prerequisite, not an afterthought.
For Microsoft, a successful 50,000-person deployment in a high-stakes environment would be a powerful case study. For IT leaders, the message is unambiguous: the winning formula for enterprise AI isn’t about the tech—it’s about the people and processes you wrap around it.