Unlock Digest Editor for free
FT Editor Roula Khalaf picks her favourite stories in this weekly newsletter.
Google is being investigated by Europe’s privacy watchdog over how it handled personal data during the development of one of its artificial intelligence models, as regulators step up scrutiny of Big Tech’s AI ambitions.
Ireland’s Data Protection Commission, the body responsible for enforcing the European Union’s General Data Protection Regulation (GDPR), said it has opened a statutory investigation into the tech giant’s Pathways Language Model 2, or PaLM 2.
PaLM 2 is due out in May 2023 and will debut before Google’s latest Gemini models, which power its AI products. Gemini, is has been released in December of the same year, is now the core model behind the company’s text and image generation services.
The investigation will assess whether the company breached its obligations under the GDPR regarding the processing of personal data of EU and European Economic Area citizens.
Under this framework, companies must conduct a data protection impact assessment before they start processing that information where the nature of how that information is used is likely to create a high risk to the rights and freedoms of individuals.
This applies particularly to new technologies and is “of vital importance in ensuring that individuals’ fundamental rights and freedoms are fully considered and protected,” the regulator said in a statement.
The review is under investigation. Google did not immediately respond to a request for comment.
This is the latest in a series of DPC actions against large tech companies that build large language models.
In June, Meta paused plans to train its Llama model on public content shared by adults on Facebook and Instagram across Europe, following discussions with Irish regulators. Meta then limited availability some of its AI products to users in the region.
A month later, X users discovered they had been “opted in” to have their posts on the site used to train systems at Elon Musk’s xAI startup.
The platform has paused processing. European User Data in August to train its Grok AI model, following legal proceedings by the DPC. It was the first time the regulator had used its powers to take such action against a tech company.
