August 27, 2025 - NASA and IBM have jointly released Surya, an open-source artificial intelligence model capable of forecasting solar flares up to two hours before they occur. The breakthrough system, trained on nine years of satellite data, demonstrates 16% higher detection accuracy than previous methods and has been made freely available on Hugging Face for global scientific collaboration.
Solar flares pose significant risks to satellite communications, power grids, and astronauts in space, making accurate prediction crucial for modern technological infrastructure. Surya utilises advanced machine learning algorithms to analyse patterns in solar magnetic field data, identifying precursor signals that human analysts might miss. "This represents a quantum leap in our ability to protect critical infrastructure from space weather events," explained NASA's Space Weather Prediction Center director in a recent briefing on the system's capabilities.
The decision to release Surya as open-source software reflects a growing trend towards collaborative AI development in scientific research. By making the model freely available, NASA and IBM enable researchers worldwide to contribute improvements and adaptations, potentially accelerating advances in space weather prediction and related fields.
Our view: NASA's open-source approach to Surya exemplifies how AI can serve the global public good when developed transparently and collaboratively. This model not only addresses a critical safety need but also demonstrates the value of government-industry partnerships in AI research. The 16% improvement in accuracy might seem modest, but in space weather prediction, such gains translate to invaluable additional preparation time for protecting billions of pounds worth of infrastructure and human lives.
beFirstComment