Bard’s big expansion will not reach Europe for now. The reason for this is likely to be privacy concerns, which may be even more profound at Google than at OpenAI.
At the I/O developer conference, Google announced several new features for its ChatGPT competitor Bard. These include support for the new PaLM 2 language model and related improved code and logic capabilities, as well as an expansion to 180 countries and territories in the three languages US English, Japanese, and Korean. All this without a waiting list.
For now, no Google Bard in Europe
However, the list of supported countries excludes countries in the European Union. “We’ll gradually expand to more countries and territories in a way that is consistent with local regulations and our AI principles,” Google writes on its support page.
Google has not yet commented in detail, but the issue is clear: European data protection authorities are currently examining the legitimacy of AI chatbots in the context of the GDPR. This concerns the training data, as well as the data entered into and generated by the systems.
ChatGPT in particular got into trouble with the Italian data protection authority, which even forced OpenAI to temporarily shut down ChatGPT. OpenAI made some concessions and now offers, among other things, an opt-out so that the data entered will no longer be used for AI training. ChatGPT is now available again in Italy, but OpenAI is still under investigation.
In Germany, state data protection commissioners initiated proceedings against OpenAI and ChatGPT in mid-April. “If personal data is used, even as training data for AI, there has to be a legal basis,” Dieter Kugelmann, head of the German states’ AI task force, told Tagesspiegel. “We need to know where the data comes from.”
Businesses also tend to be critical of ChatGPT and the like, partly because of the transfer of data to third-party servers where it could be stored, and partly because of the potential use of data for AI training. Microsoft and OpenAI plan to offer more privacy-friendly business solutions.
Google’s Bard and OpenAI’s ChatGPT: Same but different
So, it’s possible that Google is being tactical, deliberately letting OpenAI take the lead in the EU to clear up legal uncertainties. But as long as ChatGPT remains available, the user base will grow. For Google, it is playing with fire because it is already behind.
But it has more to lose than OpenAI, because it is likely to be judged much more harshly on privacy. In addition to chat data, Google has a lot of user data from other sources that it could link to chat services. For example, users could link their Google accounts to Bard, making Bard’s responses more personalized. However, this linking of services could be problematic from a privacy and antitrust perspective.
Meta, which wanted to merge social media data with VR data, had a similar experience and ran into trouble with Germany’s Federal Cartel Office. As a result, its Quest 2 VR headset, which was released internationally in October 2020, wasn’t officially launched in Germany until late 2022.