Google offers 200 European researchers AI tools and $3 million in credits

MUMBAI: Google is putting more computing power behind European science, but the growing reach of artificial intelligence is also sharpening calls for tighter controls. The company will offer up to 200 European scientific researchers access to its advanced AI models and $3 million in computing credits, Reuters reported.

Announced on Wednesday, the initiative will be delivered through Google DeepMind and Google Cloud. The European Commission will select participants under its Resource for AI Science in Europe (RAISE) programme.

The package includes five scientific AI tools: AlphaFold 3, WeatherNext, AlphaGenome, AlphaEvolve and AlphaEarth Foundations. Each is designed to support research across different fields, from understanding biological structures to forecasting weather and studying the planet.

AlphaFold 3 supports protein research, while WeatherNext focuses on weather forecasting. Google says the tools could help scientists tackle complex problems faster, including predicting extreme weather events and mapping molecular structures.

The initiative comes as advanced AI systems attract growing attention not just for their research potential, but also for the risks they may pose when safeguards fall short.

Around 400 researchers from 34 countries, including AI pioneers Geoffrey Hinton and Yoshua Bengio, called on Tuesday for an international body to oversee containment standards and the reporting of security incidents involving advanced AI models, Reuters reported.

In an open letter, the researchers proposed industry-wide standards to prevent AI systems with significant cyber capabilities from accessing external networks without adequate safeguards. They also called for a registry of major security incidents, which they said should eventually be made public.

The signatories urged organisations developing or testing advanced models to introduce stringent containment measures immediately, while governments work towards a formal oversight framework.

The researchers argued that frontier AI models should initially be isolated from external networks and placed under independent oversight. In their view, voluntary commitments from technology companies alone are not enough to manage the risks.

Their concerns follow reported security incidents in which AI models bypassed inadequately isolated testing environments and compromised external systems.

The two developments highlight the challenge facing the AI industry: expanding access to powerful tools that could accelerate scientific breakthroughs while ensuring those same systems are developed and tested responsibly.

Google’s initiative aims to put advanced AI capabilities in the hands of European researchers, while the researchers’ appeal underscores the need for stronger safeguards as the technology becomes more capable.