NIS2 compliance and multilingual documentation
In our article, we explore why NIS2 compliance is not only a matter of cybersecurity controls, but also of legal and terminological consistency across languages. For cross-border organisations, multilingual documentation plays a direct role in incident reporting, internal alignment, regulatory defensibility and risk management.
Cybersecurity regulation and legal language
The EU is currently accelerating its cybersecurity agenda with instruments like the NIS2 Directive (Directive (EU) 2022/2555) and the Cyber Resilience Act (Regulation (EU) 2024/2847).
Translation as Evidentiary Infrastructure
Translation in cross-border disputes does more than make evidence readable. It shapes what a court or tribunal recognises as “the facts”. From EU consumer claims to investment arbitration, every terminological choice can influence admissibility, liability and even access to justice itself. In our latest article, Translation as Evidentiary Infrastructure, we explore why multilingual legal proceedings require more than linguistic accuracy: they require a clear understanding of how translation actively constructs legal reality.
Language and compliance in the EU digital regulatory framework
The EU’s new digital regulations set out obligations that affect how technology companies operating in the Union handle language, documentation, and compliance.
Findings of the WMT 2024 Shared Task on Chat Translation
New findings from the WMT Chat Translation Shared Task show how AI performs in real-time, bilingual customer-support chats. The evaluation — combining neural metrics, MQM-based LLM analysis, and human reviewers — highlights progress yet clear limitations. A timely reminder that expert human judgement remains essential for meaning-preserving translation.
GPT-4 vs. Human Translators: A Comprehensive Evaluation of Translation Quality Across Languages, Domains, and Expertise Levels
A recent peer-reviewed study has offered one of the most rigorous comparisons to date between state-of-the-art machine translation and LLM engines and human translations across languages, domains, and expertise levels. Its findings highlight both the strengths and persistent limitations of AI-based translation and machine translation.
The Verification-Value Paradox: A Normative Critique of Gen AI Use in Legal Practice
Joshua Yuvaraj argues that generative AI’s persistent hallucinations and opacity make its outputs inherently unreliable for legal work. His review of global cases shows that unverified AI content continues to appear in court filings, with real disciplinary consequences. On balance, any efficiencies gained through AI would thus be outweighed by the line-by-line human verification required by responsible legal practice.
Mistranslation of Regulations and the Limits of Retroactive Correction
A single mistranslated phrase in EU legislation changed who could receive millions in State aid in Romania. Years later, the funds were clawed back—triggering legal battles and raising a fundamental question: can corrected regulations apply retroactively?
Advocate General Szpunar’s recent opinion before the CJEU sets a strong precedent: citizens cannot be expected to cross-check EU laws in multiple languages. For translators, lawyers, and policymakers, the case is a powerful reminder—accurate legal translation is not just about words, but about fairness, certainty, and economic impact.
Alleged Mistranslation Triggers International Incident
A translation of Archbishop Georgios’s Easter message by the Turkish News Agency Cyprus (TAK) was deemed misleading and “conducive to hate speech.” The case highlights the ethical sensitivites of translating conflict-related discourse.