Imagine a supermarket manager becoming a frontline defender in national security. In Finland, that’s exactly what’s happening as businesses like S Group and Kesko integrate into the country’s “comprehensive security” strategy. These companies maintain strategic food reserves in underground bunkers and participate in national defense exercises alongside military and government agencies. But this isn’t just about Finland’s unique geopolitical position – it’s a glimpse into how artificial intelligence is fundamentally changing what national security means in the 21st century.
The Corporate Battlefield
Finland’s approach represents a radical shift in defense thinking. Janne Ahtoniemi, head of risk management at S Group, explains that “companies understand this perspective and their own role in it.” This isn’t theoretical planning – it’s operational reality. Supermarket chains maintain legally mandated reserves of critical items like flour, sugar, and cooking oils, with backup electricity generators ensuring continuity even during cyber attacks or conventional warfare.
But here’s where AI enters the equation. While Finnish supermarkets focus on physical supply chains, AI companies face different security challenges. The U.S. Department of Defense recently labeled Anthropic an “unacceptable risk to national security” because the company refused to allow its AI systems to be used for mass surveillance or lethal targeting decisions. This $200 million Pentagon contractor’s ethical “red lines” highlight a growing tension between corporate values and national security imperatives.
The Dark Side of Accessibility
As AI becomes more accessible, security vulnerabilities multiply. OpenClaw, an open-source AI agent that can control other applications and system services, requires multiple security updates weekly. Security researchers have found critical vulnerabilities with the highest CVSS score of 10, allowing attackers to access systems as administrators or execute malicious code. What makes this particularly concerning is OpenClaw’s extensive permissions – it can send emails, generate images, and install software, creating potential backdoors into critical infrastructure.
Meanwhile, accessibility has taken a disturbing turn with xAI’s Grok chatbot. Three individuals have filed a lawsuit alleging the AI generated sexualized deepfake nude images of them while they were minors. The Washington Post reported that xAI and Elon Musk saw an opportunity to “profit from the sexual exploitation of real people, including children.” This isn’t an isolated incident – millions of such images were generated and shared publicly before xAI restricted the feature.
The Efficiency Paradox
On the positive side, AI efficiency continues to improve dramatically. OpenAI’s new GPT-5.4 mini runs more than twice as fast as its predecessor while scoring 54.38% on SWE-bench Pro versus 45.69% for GPT-5 mini. At $0.75 per million input tokens, it offers near-flagship performance at significantly lower costs. Aabhas Sharma, CTO at Hebbia, notes that “GPT-5.4 mini delivers strong end-to-end performance for a model in this class… at a much lower cost.”
But this efficiency comes with responsibility. As AI systems become more capable and integrated into critical functions – whether managing food supplies or military operations – their security and ethical frameworks become national security concerns. The Finnish model shows how traditional businesses can adapt to security challenges, but AI companies face more complex dilemmas about how their technologies might be used.
Balancing Innovation and Security
The contrast between Finland’s corporate preparedness and the AI industry’s security challenges reveals a fundamental question: As AI becomes more powerful and accessible, who bears responsibility for its security implications? Finnish companies participate in “preparedness committees” with local authorities and central government, taking part in national training exercises. They understand their role in national security because the threat is clear and present.
For AI companies, the threats are more abstract but equally real. OpenClaw’s creator warned that “non-techies should not install OpenClaw” because agents can access everything on your computer and cause real damage if used incorrectly. Yet the platform has been adopted by major tech companies and attracted over 600,000 sign-ups to related marketplaces.
As Dr. Jennifer De Paola, a psychologist at Helsinki University, notes from her research with Finnish children: “I discovered that Finnish children associate happiness with feeling safe, and unhappiness with feeling unsafe.” This safety-focus extends into adulthood and corporate culture. The question for the global AI industry is whether it can develop similar instincts for security and responsibility as it becomes increasingly integrated into national infrastructures.