Google’s AI Chatbot ‘Bard’ EU Launch delayed due to privacy concerns
30th June 2023
Google’s AI Chatbot ‘Bard’ EU Launch delayed due to privacy concerns
Google’s expansion of its AI chatbot Bard to 180 countries has hit a roadblock in the European Union (EU).
The Irish Data Protection Commission (DPC) has raised privacy concerns and demanded further information from Google before allowing the launch of Bard in the EU. This delay highlights the ongoing challenges faced by AI technology and chatbots in particular in complying with strict EU privacy regulations.
The Irish DPC, responsible for regulating Google’s EU operations, raised concerns about the privacy practices of Bard.
According to Deputy Commissioner Graham Doyle, Google failed to provide the necessary documentation, including a data protection impact assessment required under EU privacy laws. As a result, Bard’s launch, which was due to happen mid-June 2023 in the EU, has been postponed.
The DPC continues to examine Bard and will share relevant information with other EU privacy regulators as the investigation progresses.
Google intends to make Bard available in the EU responsibly, following engagement with experts, regulators, and policymakers. The company has been in discussions with privacy regulators to address their questions and receive feedback. However, the specific concerns raised by the DPC regarding Bard remain unclear at this time.
The EU has been closely scrutinizing AI chatbots to ensure compliance with the GDPR. OpenAI’s ChatGPT, a competitor to Bard, has already faced investigations in Italy, Germany, France, and Spain, with a pan-EU task force monitoring the situation.
Privacy concerns related to AI chatbots include inadequate protections for minors and the inability to opt out of data collection, which powers these systems. Users are often unaware that their conversations with chatbots are recorded and can be examined by human moderators.
With Bard’s delayed launch, alternatives such as Bing AI and ChatGPT remain available in the EU. Microsoft’s Bing AI, powered by OpenAI’s GPT-4, has successfully entered the European market, albeit under ongoing scrutiny. The postponement of Bard’s launch reflects the challenges companies face in navigating the EU’s privacy regulations while introducing AI chatbot technologies.
Google Bard has been available in the UK since March 2023.
A note about AI and how to ensure you are using it lawfully in your organisation
With the increasing prevalence of Artificial Intelligence, it is essential for organisations to develop a thorough understanding of how personal data should be lawfully and appropriately used in the context of AI.
While AI can offer significant advantages to organisations, it can also give rise to privacy concerns.
The ICO issued guidance in March 2023 that addresses many of these concerns and offers valuable insights into applying personal data protection laws within the realm of AI.
If you haven’t already done so and you are using AI in your organisation, you can find the ICO Guidance on AI here.
The ICO Guidance provides practical recommendations for improving the handling of AI and the associated personal information.
A summary of the key points to consider from this ICO Guidance on AI:
- Adopt a risk-based approach when developing and deploying AI.
- Provide explanations of AI decisions to the individuals affected.
- Minimise the collection of data.
- Reduce the risks of bias and discrimination.
- Properly prepare data by dedicating sufficient time and resources.
- Ensure the security of AI systems.
- Incorporate human review of decisions.
- Engage with external suppliers.
What does this mean for organisations using AI?
Organisations need to establish proper AI governance to understand if and how AI is being utilised, determine if it involves “solely automated decision-making,” and identify the associated legal risks, particularly in terms of data privacy.
Organisations employing AI should consider generating AI Explainability Statements and/or conducting Data Protection Impact Assessments (DPIAs).
Organisations should maintain robust policies and practices concerning data collection, retention, and processing, including for AI applications.
If there is a risk of discriminatory outcomes from AI systems, you must take adequate measures to mitigate these risks.
Footnote. Never heard of Google’s AI Chatbot Bard of ChatGPT?
They are chatbots that use natural language models and machine learning to answer questions and solve problems. That is only scratching the surface of what they can do now . .the question is . . what will AI be able to do next?
As always, if you would like some advice regarding data protection in your organisation, please do take advantage of your complimentary thirty minute consultation with one of our specialists.
