Public Attitudes and Security Perceptions in Bulgaria: Resilience and Vulnerabilities in 2025

Published Monday 15 December 2025 at 11:34

Bulgaria remains a key member of the Western institutional system, positioned on NATO’s southeastern flank and serving as the Alliance’s gateway to the Black Sea. With Russia’s Black Sea Fleet significantly weakened following repeated Ukrainian strikes, the strategic importance of Bulgaria’s territory and infrastructure has only increased. Yet, while Bulgaria’s formal alignment with the EU and NATO is clear, public attitudes continue to reveal deep ambivalences.

GLOBSEC Trends 2025 data show that Bulgarian respondents remain divided in their geopolitical outlook and broadly disillusioned with governance. Despite some political stabilisation after the formation of a technocratic cabinet in late 2024, trust in institutions and satisfaction with democracy remain among the lowest in CEE, reflecting years of recurring elections and fragile coalitions that have fuelled public fatigue. This discontent extends to perceptions of the EU and NATO, which many view as distant or unresponsive to domestic concerns, a narrative that is often amplified and exploited.

The latest data from GLOBSEC Trends 2025 show that, while divisions remain pronounced, public opinion in Bulgaria reflects both resilience and vulnerabilities in the country’s democratic and security landscape.

Read more in the full report here

Jana Kazaz

Large Language Models (LLMs) can generate human-like disinformation, yet their ability to personalise such content across languages and demographics remains underexplored. This study presents the first large-scale, multilingual analysis of persona-targeted disinformation generation by LLMs. Employing a red teaming methodology, we prompt eight state-of-the-art LLMs with 324 false narratives and 150 demographic personas (combinations of country, generation, and political orientation) across four languages–English, Russian, Portuguese, and Hindi–resulting in AI-TRAITS, a comprehensive dataset of 1.6 million personalised disinformation texts. Results show that the use of even simple personalisation prompts significantly increases the likelihood of jailbreaks across all studied LLMs, up to 10 percentage points, and alters linguistic and rhetorical patterns that enhance narrative persuasiveness. Models such as Grok and GPT exhibited jailbreak rates and personalisation scores both exceeding 85%. These insights expose critical vulnerabilities in current state-of-the-art LLMs and offer a foundation for improving safety alignment and detection strategies in multilingual and cross-demographic contexts. [+]

Public Attitudes and Security Perceptions in Bulgaria: Resilience and Vulnerabilities in 2025

BROD

EDMO Network

Follow us