National Petition: For a Secure Mauritian
Internet and Sovereign AI

We, the undersigned citizens and stakeholders of the Republic of Mauritius, recognize the government's commitment to transforming our nation into a high-tech, innovation-driven economy as evidenced by the FAIR AI framework. The FAIR AI framework has been announced by the ministry of ICT to provide a framework on the use of AI in Mauritius. However, we note with grave concern that while AI adoption is accelerating, the sophistication of AI-powered cyber attacks is causing a challenge for all countries including Mauritius especially with the emergence of capable AI cybersecurity models like Anthropic Mythos.

Mauritius is a regional financial and ICT hub. This status makes us a primary target for next-generation cyber-attacks. Current frameworks under the Cybersecurity and Cybercrime Act 2021 are robust for traditional threats, but AI-driven attacks operate at speeds and scales that require autonomous, AI-integrated defense systems. Reliance on existing traditional defences is no longer sufficient to protect our national critical infrastructure and data sovereignty.

With the release of Anthropic Mythos, where a 27 years old security vulnerability was found in OpenBSD, the most secure operating system on the planet. AI cybersecurity models present new challenges. Models such as Anthropic Mythos can discover security vulnerabilities at scale. Several security issues (271) were found in firefox, a well known web browser. We emphasize the need to act with urgency as the threat is real.

Cyberstorm.mu, a local R&D tech community in Mauritius, has obtained access to an AI cybersecurity model after going KYC process and was able to assess its potential to discover security vulnerabilities in OpenSSH, a widely used tool for secure management in the global IT industry. Cyberstorm.mu developed a security patch which was merged in an upcoming OpenSSH release (view on GitHub)

However, it is not clear whether cyberstorm.mu, or anybody in Mauritius is acting within the FAIR AI framework without an explicit section within the FAIR AI framework documenting this practice.

We respectfully petition the Government of Mauritius to:

• Allocate Enhanced Funding: increase the cybersecurity budget in the upcoming 2026/2027 fiscal year, specifically earmarked for Defensive AI research and infrastructure.
• Establish a new National AI Security Command (NAISC): Create a specialized, autonomous body under the Ministry of IT, Communication and Innovation dedicated solely to:
  • Detecting and neutralizing AI-generated threats in real-time.
  • Auditing public and private sector AI deployments for security vulnerabilities.
  • Developing "Sovereign AI" defensive models tailored to the Mauritian digital ecosystem.
• Mandate AI-Risk Assessments: Update the current legal framework to require all Critical Information Infrastructure (CII) providers to undergo specific "AI-Threat Readiness" certifications.
• Invest in Human Capital: Fund specialized scholarships and offensive/defensive training programs for Mauritian youth to become experts in AI-driven offensive and defensive cybersecurity.
• Improve the FAIR AI framework to recognize research for discovering and responsible disclosure of security vulnerabilities using AI cybersecurity models.

• Cyberstorm Mauritius, Cyberstorm.mu
• Linux User Group of Mauritius, lugm.org
• University of Mauritius Resilient and Innovative Computing Research Group.
• Hans Balgobin, policydraft.org