As artificial intelligence (AI) becomes embedded in critical infrastructure, government services and enterprise operations, the question of how to secure increasingly autonomous AI systems is becoming as important as how quickly they can be deployed.
Nvidia is seeking to address that challenge through the newly launched Open Secure AI Alliance, an industry coalition that brings together technology and cyber security companies to develop and share open tools, models and techniques for AI security.
The initiative builds on the Linux Foundation’s Akrites initiative and work by the Open Source Security Foundation (OpenSSF), with a focus on using open technologies to identify, remediate and disclose vulnerabilities. Nvidia argues that cyber defenders need access to AI systems they can inspect, modify and deploy themselves rather than relying exclusively on proprietary platforms.
That argument could carry particular weight in the Middle East, where countries including the UAE and Saudi Arabia are investing heavily in sovereign AI capabilities while simultaneously strengthening national cyber security frameworks. Abu Dhabi-based G42 is among the companies joining the alliance, alongside Microsoft, IBM, Cisco, CrowdStrike, Cloudflare, Dell Technologies, Hugging Face, Palantir and the Linux Foundation.
Moving AI security into the open
The alliance reflects a broader debate over whether the most powerful AI systems should remain closed and controlled by individual technology providers or whether greater openness can make them safer.
Nvidia’s position is that the industry needs both. In cyber security specifically, it argues that open models and tools allow defenders to examine how systems work, adapt them to their own environments and operate them on infrastructure they control. This could also reduce dependence on individual vendors and avoid creating single points of failure.
Irene Corpuz, founding partner of Women in Cyber Security Middle East, believes the approach could change how enterprises think about protecting AI. “It could shift AI security from relying on proprietary solutions to adopting open, transparent and collaborative security tools,” she told Computer Weekly. “This would enable organisations to better audit, customise and validate AI systems while becoming more resilient through shared standards and community-driven innovation.”
Corpuz also sees a need for governance around the alliance itself, including a common code or set of ethical principles designed to protect its integrity.
The model is familiar to traditional cyber security. Open source technologies have long allowed researchers and defenders to examine vulnerabilities collectively, develop fixes and share threat intelligence. Applying that philosophy to AI, however, introduces new complexity because increasingly capable models and autonomous agents can potentially be used by both defenders and attackers.
A sovereignty question for the Gulf
For the Middle East, openness is not only a cyber security issue. It intersects with a wider push towards technological and data sovereignty.
Governments across the Gulf are investing in domestic AI infrastructure and models while seeking greater control over where sensitive information is processed and stored. The ability to deploy security models locally could therefore become increasingly important for government agencies, critical infrastructure operators and companies handling sensitive data.
“For the UAE and Saudi Arabia, it could accelerate secure AI adoption by supporting digital sovereignty, enabling local deployment of trusted AI, and strengthening national cyber resilience in line with their AI and cyber security strategies and data residency,” said Corpuz.
Saudi Arabia provides a particularly timely example. The National Cyber Security Authority is currently consulting on new AI Cyber Security Guidelines designed to ensure cyber security requirements are applied to AI systems and to mitigate the risks associated with the technology.
The proposed framework covers four areas: cyber security governance, cyber security defence, cyber resilience and third-party cyber security. Meanwhile, the Saudi Data and Artificial Intelligence Authority’s national strategy positions responsible AI adoption, advanced digital infrastructure and cyber security resilience among the Kingdom’s strategic priorities.
G42’s participation in the Open Secure AI Alliance reflects the UAE’s broader strategy to develop sovereign AI capabilities while engaging in global debates around open source model security. The Abu Dhabi-based technology group has positioned open frontier models as essential to retaining strategic control over how AI systems are built and deployed domestically.
For organisations in both markets, the potential attraction is the ability to customise defensive AI around local infrastructure and regulatory requirements rather than sending sensitive security data to externally controlled systems.
When open AI becomes a weapon
Greater openness nevertheless creates a security dilemma. The same accessibility that enables defenders to inspect and modify AI systems can potentially allow malicious actors to do the same.
Nvidia acknowledges that open models can be misused, including by weakening safeguards or adapting capabilities for cyber attacks, although it argues that such risks are not exclusive to open systems. Corpuz similarly cautioned that organisations should not interpret “open” as inherently secure.
“Yes, open tools can be exploited by attackers, but they also empower defenders,” she said. “The key is to implement strong governance, secure-by-design practices, continuous testing, access controls most especially, and threat monitoring to ensure the benefits outweigh the risks.”
That tension has become more significant as AI evolves from tools that primarily generate content towards agents capable of executing actions autonomously. The significance of the Open Secure AI Alliance will ultimately depend on whether it can turn industry collaboration into tools and standards that enterprises can deploy.
For Middle Eastern organisations, its progress will be particularly relevant if open security frameworks can be combined with sovereign infrastructure, local data requirements and national cyber security controls.
The UAE and Saudi Arabia have spent the past several years building AI infrastructure and encouraging adoption across government and industry. As those deployments mature, the focus is increasingly moving from access to AI towards questions of control, resilience and trust.
The Open Secure AI Alliance presents one possible answer: rather than concentrating AI defence inside a small number of proprietary systems, give organisations the ability to inspect, adapt and operate more of the security layer themselves.
But openness alone will not solve the AI security problem; as Corpuz explained, the value of open defensive tools will depend on the governance, access controls and continuous monitoring surrounding them.
For Gulf organisations balancing rapid AI adoption with national requirements around cyber security and sovereignty, that may prove to be the more consequential debate.
