G42 Unveils Comprehensive Frontier AI Safety Framework To Ensure Responsible Development

G42, a technology group based in the UAE, has unveiled its Frontier AI Safety Framework. This initiative supports the AI Seoul Summit’s Frontier AI Safety Commitments and the Bletchley Declaration. The framework outlines protocols for risk assessment, governance, and external oversight to ensure advanced AI models are developed safely and responsibly.

The framework introduces a multi-layered approach to managing AI risks. It ensures that advanced AI systems are developed, tested, and deployed with responsibility. This includes establishing frontier capability thresholds, independent governance mechanisms, and deployment safeguards to identify and mitigate potential AI risks before they become critical.

The G42 Frontier AI Governance Board plays a crucial role in overseeing model compliance and safety protocols. Led by Dr. Andrew Jackson as Chief Responsible AI Officer, the board includes Alexander Trafton as Head of Technology Risk, Martin Edelman as General Counsel, and Adele O’Herlihy as Head of Responsible AI. They manage incident response and ensure adherence to safety measures.

Peng Xiao, Group CEO of G42, stated: "AI is the defining technology of our era—an essential utility that will reshape economies and societies, much like electricity did in the past." He emphasised that with such power comes responsibility, highlighting the framework's role in ensuring innovation proceeds with appropriate safeguards.

To maintain transparency and compliance, G42 will conduct internal governance audits alongside annual external reviews. The company plans to publish a transparency report sharing key safety insights and risk assessments. This ensures accountability in their AI development processes.

The framework also sets clear risk thresholds for assessing biological threats, cybersecurity vulnerabilities, and autonomous decision-making risks. If a model nears these thresholds, G42 will implement additional safeguards or modify system behaviours to limit deployment.

Collaborations and Global Engagement

G42 is committed to collaborating with regulators, policymakers, and industry partners to enhance AI safety practices globally. The company aims to contribute to global governance discussions by engaging with other signatories of the commitments.

The development of this framework was supported by leading AI risk experts such as METR and SaferAI. Their insights were instrumental in shaping governance strategies outlined within the framework.

X-Risks Leaderboard Initiative

To operationalise its safety framework beyond policy discussions into measurable actions, G42 launched the X-Risks Leaderboard. This open evaluation platform assesses AI model risks in cybersecurity, chemistry, and biology using G42’s Safety Evaluation Suite at Inception.

G42 continues building on partnerships with companies like Microsoft, NVIDIA, AMD, Cerebras, and Qualcomm. By sharing threat intelligence with industry partners through these collaborations they address common challenges while tackling emerging risks effectively.

As one of the first Middle Eastern firms introducing comprehensive AI safety measures like this framework demonstrates leadership in both governance aspects related specifically towards mitigating associated risks involved when dealing with artificial intelligence advancements today.

With inputs from WAM