Burhanettin Duran, Head of Communications for the Turkish Presidency, outlined a new artificial intelligence ecosystem built entirely on domestic infrastructure to secure institutional data. This shift gives public sector communicators direct control over generative tools, avoiding third-party data leaks while improving media monitoring and content production.
Securing institutional data
The Directorate of Communications deployed air-gapped large language models (LLMs) and vision language models (VLMs) hosted on in-house GPU servers. Duran said in a recent social media post: "We built an autonomous artificial intelligence architecture using our own GPU servers and infrastructure, and our data remained within the institution and was protected." This setup ensures sensitive information never leaves the organization.
To support internal workflows, the directorate launched CibGPT, a closed-loop AI platform restricted to staff, and Δ°lgen, an AI coding agent designed to speed up software development. These tools form the foundation of a broader AI for Government strategy that prioritizes technological independence.
Automating media and crisis analysis
For communications teams, the most immediate applications involve processing large volumes of public information. The directorate integrated AI into its international media monitoring to increase the speed and accuracy of risk, crisis, and agenda analysis.
The team also adopted VLM-based video generation systems to handle multilingual content production at high volumes. Additionally, the directorate connected natural language interfaces to legacy databases through a system called ULAK, allowing non-technical staff to query historical records without coding knowledge.
Why this matters for PR and communications
Public sector communicators face strict data governance requirements that often block the use of commercial AI tools. By adopting air-gapped, institution-specific models, PR teams can safely automate media monitoring, crisis detection, and multilingual content creation. Professionals tracking AI for PR & Communications should note this shift toward closed-loop systems, as it proves that sensitive institutional data can be processed by generative models without exposing it to external vendors.
Your membership also unlocks:
