Telecom Giants Launch Open AI Initiative
The GSMA, the global telecoms industry body, has launched a new open-source initiative called “Open Telco AI.” The effort aims to standardize and accelerate the development of AI for network automation, customer service, and data analytics across the sector.
General-purpose AI models often struggle with the specialized language and data of telecommunications, with tests showing they can provide 30-40% incorrect responses on specific technical topics like 3GPP standards. This initiative addresses a significant gap, as only 16% of generative AI deployments in the telecom sector have been applied to network operations. Founding supporters AT&T and AMD are providing core components. AT&T is releasing a family of open, hardware-agnostic AI models trained on public data, while AMD is supplying the necessary computing power for training and testing through its GPU platforms and cloud partner TensorWave. The collaboration extends globally, with participation from major operators like China Mobile, China Telecom, China Unicom, Deutsche Telekom, Telefónica, and Vodafone. Tech giants including Google Cloud and IBM are also involved, contributing to a broad industry-wide effort. Progress will be measured by a new "Telco Capability Index," a benchmark for tracking AI model performance on a range of telecom-specific tasks. The goal is to create a shared, open foundation of models, data, and tools to build AI that truly understands the complexities of network environments. This effort mirrors the GSMA's earlier "Open Gateway" initiative, which successfully created a standardized framework of APIs to give developers universal access to operator network capabilities. Both initiatives aim to foster innovation by creating common, open standards across the industry. The ultimate aim of applying AI in telecom is to move from reactive to proactive operations. Key applications include predictive maintenance to prevent network outages, dynamic traffic routing to improve service quality, and enhanced security through faster threat detection.
