The recent release of groundbreaking Generative AI models like Dall-E and ChatGPT has created significant attention from the media and the public. However, it's important to consider these technologies' potential Data Privacy implications. Organizations using Generative AI must be aware of three main areas of concern: unauthorized access and data breaches, AI transparency and derivative works.

What is Generative AI?

Generative AI is artificial intelligence that uses algorithms to generate new content based on data from various sources. These AI models can create images, like Dall-E, or text, like ChatGPT, by using large amounts of data and computing power. While there are many different Generative AI models, Dall-E and ChatGPT have gained particularly widespread attention.

Unauthorized Access and Data Breaches

Generative AI models are trained on large volumes of data, including images and text from the Internet or provided to the creators of the AI model. However, there is a risk that some of this data may not have proper consent for use. For example, I recently created a Generative AI and Data Privacy video about a woman who discovered that an image of herself, which she had only given consent to be used in a medical study, had been included in a publicly available Generative AI data set. This could have occurred due to poor security of a cloud data source or the image could have been provided to the AI model's creators without proper proof of legal consent. The use of data without proper consent, particularly sensitive data, could be evidence of unauthorized access or a data breach. Organizations handling sensitive data or images should review their data protection methods to ensure that this information is not made public in Generative AI data sets.

Transparency of AI Algorithms

There has been much discussion about the importance of transparency in AI and how these systems work. In some areas, such as the EU and the UK with the Algorithmic Transparency Standard, there are efforts to pass laws and establish standards for AI transparency. Generative AI will be no exception. Regulators are particularly concerned about AI that may cause harm to humans, such as reinforcing bias, spreading misinformation, or resulting in negative outcomes based on the generated information. As organizations begin using Generative AI, it's crucial to assess the potential negative impacts on humans and be transparent about how the technology is used.

Derivative Works

Derivative works are the resulting "work product" created from data inputs to digital systems. When using Generative AI, some people may voluntarily submit images or text to AI models to receive a new image or revised text based on the model's calculations. However, these users may not realize they are giving away the rights to use their text and images to the Generative AI product company. While this may be acceptable to some people, it's important to understand that providing text and images to AI models means that the organization may use this data in any way they choose at any time in the future. Organizations should consider the potential risks of adding images and text to Generative AI systems if they want to retain control over their intellectual property.

Generative AI has the potential to create new and exciting content, but organizations need to be aware of the Data Privacy implications. Careful consideration of unauthorized access and data breaches, AI transparency, and derivative works can help ensure that the privacy of users and customers is protected as these technologies are adopted and that organizations can make Data Privacy a Business Advantage.