US regulator investigating ChatGPT over bad content

 

The US Federal Trade Commission is investigating OpenAI to determine if its hugely popular ChatGPT app harms consumers by generating false information and whether its technology mishandles user data.

Microsoft-backed OpenAI was notified of the investigation in a 20-page questionnaire in which the company is asked to describe incidents in which users were falsely disparaged, and share any company efforts to ensure this does not happen again.

The investigation by the US regulator was first reported by The Washington Post. OpenAI's release of ChatGPT last November stunned the world as it displayed the power of large language models (or LLM), a form of artificial intelligence known as generative AI that can churn out human-like content in just seconds.

Amid the marvel at the technology's capacities, reports came in that the models could also churn out offensive, false or just strange content, sometimes called "hallucinations".

FTC chair Lina Khan addressed a congressional committee hearing on Wednesday, and while she did not mention the investigation, she told lawmakers that her agency had concerns about ChatGPT's potentially libellous output.

"We've heard about reports where people's sensitive information is showing up in response to an inquiry from somebody else," Khan said.

"We’ve heard about libel, defamatory statements, flatly untrue things that are emerging. That’s the type of fraud and deception that we are concerned about," she added.

The FTC's investigation is mainly focused on how this aspect could harm users, according to the questionnaire, but also delves into OpenAI's use of private data to build its world-leading model.

The company's GPT-4 is the bedrock technology behind its own ChatGPT as well scores of other programs from companies that pay a fee to OpenAI to access its model for their own uses.

An FTC probe does not necessarily bring further action and the regulator can close the case if it is satisfied by the target company's answer.

If the FTC perceives illegal or unsafe practices, it will demand remedial action and possibly launch a lawsuit. OpenAI and the FTC did not respond to a request for comment.