The accelerating evolution of Artificial Intelligence (AI) is redefining how nations approach innovation, governance, and global competitiveness. In recognition of the transformative impact of AI across key sectors, the Cypriot Council of Ministers has established the National Artificial Intelligence Taskforce (AI Taskforce) to lead the country’s innovation and AI readiness.
The AI Taskforce has been formed to serve in an advisory capacity, guiding the ethical and responsible integration of artificial intelligence across Cyprus. Its mission focuses on driving economic growth, strengthening the nation’s competitive edge, and delivering tangible benefits to citizens through smarter, more efficient public and private sector innovation.
Follow THE FUTURE on LinkedIn, Facebook, Instagram, X and Telegram
Among the experts appointed to the Taskforce is Sotirios Chatzis, one of Cyprus’s most distinguished AI researchers. With over two decades of experience in end-to-end multimodal generative models and a global track record in European projects, Chatzis brings a rare depth of technical and strategic knowledge to the table.
In this exclusive interview for The Future Media, Chatzis shares his perspective on Cyprus’s unique opportunities in AI, the challenges of implementation, and the future he envisions for the island as a serious player in the European AI ecosystem.
What perspectives or expertise do you bring to the Taskforce that you believe will shape Cyprus’s AI future?
I am the country’s only researcher trained (and having worked for two decades exclusively) on end-to-end multimodal generative models, which form the foundation of any (modern) generative AI system. Thus, I bring hands-on knowledge of how to design, train, and deploy these systems safely.
My team focuses on building the algorithms that sit under headline models— Mixtures of Experts (MoE) / Winner-Takes-All architectures, preference-driven reinforcement-learning (RL) fine-tuning / DPO, variational expectation-maximization (EM) training loops—and then proves them in real-world pilots, from financial-time-series reasoning to auditable autonomous driving systems.
This “full-stack” perspective (theory → infrastructure → application → policy) enables the Taskforce to have a deep understanding of how to transcend from strategic intent to working systems quickly.
In your view, what qualities or achievements made you a strong candidate for this national role at such a transformational moment in AI development?
First of all, I’m the only researcher in Cyprus to have published end-to-end generative foundational models that are now commonplace in headline AI systems. I also have a track record serving as the principal investigator of European Horizon projects, summing up to a total budget of more than €12 million, giving me direct experience with compliance and commercialization. And finally, my past affiliations with Imperial College London and the University of Miami’s generative computer vision group help connect Cyprus to top-tier labs for joint datasets and computing.
What AI-driven opportunities do you see as most transformative for Cyprus’s key sectors—tourism, energy, and healthcare—and for society as a whole?
Tourism can leapfrog with multimodal concierge agents that ingest images, voice, and location to build personalized cultural itineraries on-device, reducing cloud cost and latency.
Energy will benefit from foundation-model-powered digital twins of the grid that predict faults and optimize solar-storage dispatch, which is vital for our renewable targets.
Healthcare can deploy vision-language models that fuse MRI, EHR text, and genomics for earlier oncology diagnosis; small hospitals gain specialist-level triage without losing data sovereignty. Together, these raise productivity and exportable IP far beyond the size of our domestic market.
How can Cyprus align its AI development with global trends while overcoming challenges like a small market size to attract global investment and attention?
We need to begin by implementing an AI regulatory sandbox fully aligned with the EU AI Act timelines—e.g., the provisions on general-purpose model obligations effective from the 2nd August 2025—and market Cyprus as the fastest route to EU-wide compliance.
A national compute cooperation, so pooling university and private-sector GPUs into a national cluster that exposes sovereign-cloud APIs, can offer investors valuable access to compute, plus favorable R&D tax credits.
Finally, an open-source first mandate for publicly funded models would release checkpoints. This attracts the global developer community and follow-on venture backing.
How can Cyprus attract and retain top AI talent to drive its innovation agenda?
One critical initiative would be launching a Digital Thalassa Fellowship— a seven-year, tax-free package designed for post-doctoral researchers and engineers willing to co-found ventures with local SMEs. This not only brings talent in but integrates it within the Cypriot innovation ecosystem.
We can also create joint professorships, allowing start-ups to second senior scientists to universities part-time. This mirrors successful models like Israel’s Technion model, bringing together both academic and commercial innovation.
Lastly, we must acknowledge that talent follows quality of life as much as attractive salaries. Cyprus has a natural advantage here in terms of lifestyle and climate, but we also need to consider additional benefits such as subsidizing daycare and relocation costs.
What are the biggest risks associated with AI integration in Cyprus, and how can the Taskforce proactively address them?
A major risk is systemic bias in small-language data. The under-representation of the Greek language can inadvertently propagate discrimination across services. To mitigate this, we must fund national language corpora and enforce bias audits before any AI procurement.
Another concern is the concentration of compute and knowledge in the hands of a few providers, which could effectively hold the entire ecosystem hostage. To counter this, Cyprus should establish a public–private GPU reserve and open MLOps standards that make it easier to move between providers.
Regulatory lag is also a real threat. Many SMEs may hesitate to engage with AI due to their fear of non-compliance costs under the EU AI Act. The solution could be to offer one-stop compliance clinics and provide pre-approved templates for “risk-classification” templates to ease their worry.
What is the one thing you hope to achieve during your term on the Taskforce?
My main goal is to see the Cyprus Sovereign Multimodal Model—a 7 billion-parameter vision-language-audio LLM trained on local culture and released as open-source—running in production across ministries, hospitals, and tourist gateways.
What steps is the Taskforce taking to ensure that Cyprus’s AI strategy is resilient to international shifts in regulations and technology trends?
We are adopting a modular governance framework where policies reference principles (such as risk tiers and transparency) rather than fixed model classes. This gives us flexibility during periods of paradigm shifts.
We are establishing an AI Act observatory that updates national guidelines within 60 days of any EU delegated act.
When it comes to technology, we’re building standard interfaces like HuggingFace, MLflow, and MCP into all state-funded projects. This allows for quick vendor swap-outs when global tech trends change.
If you could implement one groundbreaking AI initiative in Cyprus today, what would it be and why?
I would deploy a pan-Cyprus medical-imaging diffusion model, trained on anonymized scans from every hospital in the country. This system would cut the radiology backlog by over 40 percent, democratize expertise to rural clinics, and generate exportable med-tech IP while still complying with EU data protection laws.
What advancements in AI do you think will most impact global industries in the next five years?
We’re currently on the edge of agentic multimodal models, such as those emerging from Gemini Robotics that can perceive, reason, and act in the physical world. These will fundamentally redefine logistics and manufacturing.
Open-source foundation models such as DeepSeek, Qwen, and Llama are already rivaling proprietary systems, commoditizing core capabilities once only accessible to tech giants.
Diffusion-based simulators are transforming how we build autonomous systems, generating synthetic sensor data that slashes annotation cost.
And finally, the community is now developing self-verifying LLMs that use internal chains-of-thought plus RL critics to guarantee factuality.
How do you see the European Union’s AI Act influencing the development and deployment of AI technologies across Europe and in Cyprus?
The Act’s risk-based approach creates a single digital market for trustworthy AI.
For Cyprus, early alignment means Cypriot start-ups can “passport” their products across the EU without needing to pass additional audits.
The staggered deadlines, starting with the prohibitions and AI-literacy duties from February 2025, the general-purpose model rules in August 2025, and regulations for high-risk verticals by 2027—all these give SMEs a clear compliance runway. It also allows our Taskforce to tailor sandbox incentives around each phase.
What does success look like for Cyprus in AI by 2030?
By 2030, I hope to see one unicorn-level AI company headquartered in Limassol. I envision public services that default to bilingual multimodal assistants, halving admin overhead. AI should account for at least 7% of our GDP.
Most importantly, I want Cyprus to be recognized as the Mediterranean’s premier AI testbed where global firms come to pilot responsible AI before scaling to the rest of the EU.
This vision is ambitious, but with focused execution and our unique expertise in generative and multimodal AI, it is fully within reach.
