Canadian Regulator Finds Violations in Early Development of ChatGPT

JAKARTA - The Canadian Privacy Commissioner along with privacy regulators in Quebec, British Columbia, and Alberta found a privacy violation against the way OpenAI trained and launched the artificial intelligence-based chatbot ChatGPT in the early stages of its development.

In the joint investigation, the four regulators assessed the company's initial practices as not being in accordance with applicable privacy laws, ranging from excessive collection of personal data, lack of valid consent and transparency, inaccuracies in personal information, to obstacles for users to access, correct, or delete their data.

"Through this investigation, we identified privacy issues that arose early on and made recommendations that led to improved protection of the personal information of Canadians," Canadian Privacy Commissioner Philippe Dufresne said in a statement on his official website.

The regulator also assessed that OpenAI does not have adequate accountability for personal information under its control.

In response to the investigation, OpenAI is said to have implemented a number of changes to improve the protection of personal data in the development and operation of ChatGPT.

In particular, the regulator said that Open AI has significantly restricted personal information and sensitive information used to train the new Chat GPT model.

They will also take steps to better inform Canadians about the implications of using Chat GPT.

Philippe Dufresne assessed that the steps that OpenAI has implemented, including additional commitments in the coming months, are able to address regulator concerns regarding compliance with PIPEDA, Canada's federal private sector privacy law.

"Adequate safeguards are the foundation of responsible innovation. I hope that the findings of this investigation will provide information and advance the design of privacy protection in other AI-based technologies," he concluded.