The temporarily-delayed Google Bard is now available in the European Union, as the Irish Data Protection Commission (DPC) has decided that Google has provided enough information to assuage EU privacy concerns.
Google Bard has been available since March in a number of countries, but hit an extended delay in Europe due to concerns raised by the strong personal data protections of the General Data Protection Regulation (GDPR). The AI tool has also now launched in Brazil, bringing the total of countries it is available in to over 230.
Google Bard clears EU roadblock
The Irish DPC has relented after a meeting with Google executives and staff, in which EU privacy concerns were apparently addressed to their satisfaction. Little has been made public about this meeting, but engineering vice president Amar Subramanya said that Google Bard is adding the ability to opt out of personal data collection. Google also reportedly provided regulators with a data protection impact assessment, which is not available to the general public.
At the moment, use of Google Bard requires logging in to either a personal or Google Workspace account. Under default settings, the AI service will store activity for up to 18 months; this can be reduced to three months by changing settings. Text entered into Google Bard may be paired with location information, IP addresses, or the home or work addresses stored in Google accounts. Users can also opt into a temporary “pause” of the saving of Bard activity, but it will always be held for at least 72 hours, something that Google claims is necessary for basic function.
There had already been a planned delay of the rollout of Google Bard due to EU privacy concerns, even before the Irish DPC got involved. The first countries to receive early access on March 21 were the United States and United Kingdom, and by May about 180 countries had access without a waitlist. But the service was not scheduled to roll out in the EU until at least June, and the regulator’s concerns ended up adding several more weeks to the process.
All of this comes amidst broader EU privacy concerns about “generative AI” and “large language models,” which have also impacted competitors such as ChatGPT. The European Data Protection Board (EDPB) is currently taking point on the issue for the bloc as a whole, in the midst of a special task force investigation meant to develop and harmonize enforcement standards across different national data protection authorities. For its own part, Google has been asked to perform a review and meet again with the Irish DPC in October.
EU privacy concerns are far from the full extent of Google’s problems
Google Bard’s EU expansion has been accompanied by a massive feature rollout, something the company hopes will no doubt improve public opinion of the embattled chatbot. New features that are coming online just in time for EU users are the support for 40 languages, a text-to-speech function, five different conversational styles, and Google Lens compatibility among others.
Some positive PR is badly needed, as EU privacy concerns are actually among the more minor of Google Bard’s problems at the moment. While the US does not have a comparable national data protection law, Google is not immune to private lawsuits and is already facing a class action over its scraping of training data from assorted websites.
This follows general concerns about the chatbot’s capability, especially as compared to the current state of main rival ChatGPT. The first reviews of Google Bard were about as bad as possible, as it was plagued with errant responses and even went so far as to cite irrelevant or entirely made-up studies as evidence. Even when it worked as intended, reviewers frequently criticized it for creating dry and uninteresting content and standardized tests demonstrated that overall performance at common tasks was inferior to that of ChatGPT. Some Google employees publicly opined that the product had come out of the oven far too early and should have had its release delayed.
And the behind-the-scenes controversy is not limited to solvable technical issues. The company is now under fire for labor issues in the development of Google Bard, with a recent Bloomberg article reporting that a small army of contractors who evaluate the accuracy and relative helpfulness of information that feeds the machine are paid poorly and overworked with inadequate instruction for the tasks they are given. Workers can also encounter violent or disturbing sexual content as part of the job. This follows reports that ChatGPT had outsourced similar tasks to Kenya and paid workers there the equivalent of $2 per hour.
Google Bard may also face the same enhanced scrutiny that ChatGPT did after it went live in Europe, something that culminated in a temporary ban by Italy (also due to EU privacy concerns) and the initiation of investigations in several other countries that are still open.