Google Has To Put Off The Launch Of Its AI Robot In The EU Because The Privacy Regulator Is Worried

According to the Irish Data Protection Commission (DPC), which is the tech giant’s main data protection body in the European Union, Google has put off this week’s launch of its generative AI chatbot, Bard. The bot was supposed to go live in the EU this week.

Politico was the first to report on the change, which happened a long time after OpenAI released a free research sample of its rival chatbot, ChatGPT, in November 2022. There were no restrictions on where in the world Internet users could access it.

Graham Doyle, the DPC’s deputy commissioner, said today that Google told the authority “recently” that it plans to start Bard in the EU “this week”. But he said that the company had not given the regulator enough information before the planned date, so the start would not happen when planned.

“At this point, the DPC had not been given a detailed briefing or seen a DPIA or any other supporting documents,” Doyle said in a statement. “It has since asked for this information as soon as possible and asked Google a few more questions about data protection. It is waiting for Google to answer these questions, so Bard will not launch this week.”

No one knows when Bard might now come out in Europe. But it’s important to note that Europeans have been able to use similar large language model (LLM) technology for free for months, since OpenAI didn’t limit access to its ChatGPT study preview. Users in the region can also easily access Google’s Bard if they use a VPN to set their address to the U.S. or another market where the tool is available.

The DPC also didn’t give any specifics about the issues it has brought up with Google about Bard.

Other EU DPAs have already found a number of data protection concerns with ChatGPT that may also apply to Google’s case. These include the legal basis for using people’s data to train LLM AI models, compliance with transparency requirements built into regional privacy laws, and how developers deal with problems like AI-generated misinformation (their creators say that all AI chatbots “hallucinate”) and addressing churn.

Doyle’s only other public statement was, “The matter is still being looked into by the DPC, and as soon as we get more answers to our questions, we will share information with our fellow DPAs.”

In April, EU DPAs decided to set up a taskforce through the European Data Protection Board to coordinate how they enforce ChatGPT. So the DPC is probably planning to add any lessons learned to that work, if and when it makes sense.

This spring, OpenAI’s rival chatbot, ChatGPT, quickly caught the attention of a number of EU data protection authorities. In April, Italy’s DPA, Garante, told ChatGPT to temporarily stop its service there after the DPA there got involved. (Unlike Google, OpenAI is not based in any EU Member State. This means that all EU DPAs are able to step in if they have issues under the EU’s General Data Protection Regulation (GDPR). However, only the Irish DPC is able to oversee Google’s chatbot.)

Helen Dixon, the commissioner of the DPC, has criticised rushed bans on generative AI chatbots in the past. In April, she asked regulatory bodies to figure out how to apply the bloc’s rules to the technology before banning it.

So it’s important to note that there is no longer a hard ban from Ireland. Instead, there is a delay that can’t be measured and is directly linked to Google’s lack of information and some vague “data protection questions.”

Unlike when the Garante got involved with ChatGPT, Europeans don’t know what problems the DPC is bringing up with Google. So there’s no way to know how big of a deal this action by the Irish regulator might be for this powerful AI tool. Or, if it could cause Google to be forced to offer the same data disclosures as OpenAI and give users more control, like ChatGPT did after the Italian job.