Airline dispatchers, grid operators, and farmers around the world rely on weather forecasts to make critical daily decisions.
While most people glance at a forecast for a moment, these predictions drive major strategic choices involving money, livelihoods, and safety. Farmers use them to select crop varieties, schedule fertiliser, and manage irrigation. Utilities rely on them to locate solar and wind farms and set wholesale electricity prices. Authorities use data to warn of extreme weather and trigger emergency responses. A newer sector, prediction markets, has also emerged, allowing people to bet on real-world events, including specific weather conditions.
However, the desire to manipulate weather data for an advantage in these markets, combined with a shift toward data-driven AI forecasting, threatens the accuracy of predictions. These risks are currently manageable but could escalate into systemic problems if left unchecked.
Accurate weather observations are collected from stations at airports, utilities, and transport services. Traditional systems like the Weather Research and Forecasting model or the European Centre for Medium-Range Weather Forecast Integrated Forecasting System combine these observations with numerical approximations to estimate future patterns.
Issues at weather stations, such as instrument failure or equipment upgrades, can be identified in real time or corrected retroactively. Traditional systems also use data assimilation, weighing incoming measurements against physical model expectations and readings from nearby stations.
These mechanisms generally keep observations reliable. Yet new threats are emerging. Earlier this year, reports indicated that the weather station at Paris Charles de Gaulle Airport (CDG) was manipulated to record suspicious temperature spikes on 6 April and 15 April 2026. Authorities suspect a hand-held hairdryer or lighter may have caused the issue. The incident resulted in significant payouts for online prediction-market gamblers who bet on a temperature of 22 °C (71.6°F) on days when the actual average was around 18°C (64.4°F). One individual won $20,000.
In this instance, human monitoring caught the anomaly. A French climate nonprofit association noticed the irregularities by chance and raised the alarm.
But what happens if no human monitoring exists? What if someone remotely nudges readings at many stations simultaneously, making each change small enough to appear plausible? Existing quality controls struggle to detect this kind of coordinated manipulation. Time also works against us; thorough checks of data and metadata take hours or days, yet forecasts must be released on schedule regardless of conditions.
The shift toward artificial intelligence in weather prediction raises the stakes. These methods depend heavily on accurate observations and are known as data-driven models. Researchers at ECMWF are exploring whether high-quality forecasts can be produced directly from raw observations, bypassing the assimilation step that currently acts as a quality filter. Other researchers are combining geospatial data with large language models and agentic AI to support real-time, autonomous decision-making during extreme events like storms.
Potential benefits include improvements in accuracy, efficiency, and speed. Removing humans from the equation, however, introduces new risks.
At the lower end of the risk scale, an individual speculator manipulates a station for personal gain, as seen at CDG Airport. A more serious scenario involves a group of traders coordinating to bias forecasts of renewable energy output, moving wholesale electricity prices and causing losses for counterparties. At the highest level, a state actor or saboteur could manipulate stations to trigger early warning systems prematurely or keep them silent when needed. The risk grows from fraud to compromised disaster preparedness and national security.
As long as financial or other incentives exist to manipulate observational data, adversaries will seek new opportunities. Three steps are required to stay ahead.
1. Watch the stations
Data quality controls must include station security, anomaly detection, correction, and human oversight. Weather stations require continuous monitoring to deter tampering. Data homogenisation methods that clean weather records need to accelerate, aiming to catch problems in real time. This becomes increasingly vital as agentic AI systems use this data for real-time decisions. Human oversight remains essential to flag questionable data and model outcomes. After all, humans identified the manipulation at CDG Airport.
2. Protect the data to safeguard the AI
Data defence mechanisms must be positioned throughout the AI pipeline. AI explainability and adversarial robustness tools can help understand underlying data and model outputs, identify data or model issues, and increase resilience against adversarial attacks.
3. Ensure continuous accountability along the chain
Observational data passes through many hands: operators running the stations, national weather services stewarding the records, and forecasting centres turning them into predictions. No single entity can protect data integrity alone. Each must guard its own link, and any anomaly needs communication across the whole chain, from station operators to those acting on the forecast.
What it means
The incident at CDG Airport should serve as a wake-up call. As observational data becomes more central to weather forecasting, the sector must adapt to evolving threats. This requires protecting data and models by strengthening existing oversight structures and improving coordination among key partners.
This op-ed was written by:
- Monique Kuglitsch — Innovation Manager at Fraunhofer Heinrich Hertz Institute and Chair of the UN Global Initiative on Resilience to Natural Hazards through AI Solutions
- Jesper Dramsch — Scientist for Machine Learning at the European Centre for Medium-Range Weather Forecasts (ECMWF), where they work on AIFS (Artificial Intelligence Forecasting System), ECMWF’s data-driven weather prediction model
- Franz G. Kuglitsch — Climate Scientist and Executive Secretary of the International Union of Geodesy and Geophysics (IUGG) at the GFZ Helmholtz Centre for Geosciences in Potsdam
- Andrea Toreti — Senior Scientist at the European Commission’s Joint Research Centre (JRC), where he coordinates the European and Global Drought Observatory under the Copernicus Emergency Management Service
