FTC Investigates OpenAI Over Chatbot 'Hallucinations'

Artificial intelligence company OpenAI, already under scrutiny in Congress, is now facing a Federal Trade Commission investigation into the large language model ChatGPT and its ability to generate falsehoods.

The agency this week sent a 20-page letter to the company, demanding information about a host of topics, including how OpenAI handles fabrications, or “hallucinations,” by the chatbot, as well as how the company obtains and vets the data used to train the tool.

The FTC's letter to OpenAI, which was posted Thursday by The Washington Post, specifically asks whether OpenAI scraped training data from the web or purchased information from others, as well as how the company selected the training data.

Other questions focus on how OpenAI handles statements by ChatGPT about individuals -- including potentially defamatory statements.

“Describe in detail the extent to which you have taken steps to address or mitigate risks that your large language model products could generate statements about real individuals that are false, misleading or disparaging,” the FTC asks.

The FTC also posed some privacy-related questions -- including how OpenAI handles personal information in the data, and whether ChatGPT can “generate statements about individuals containing real, accurate personal information.”

The inquiry comes as a growing chorus of voices -- including lawmakers, consumer advocates and at least one business group -- are pushing for regulations governing artificial intelligence. OpenAI has also been hit with lawsuits over copyright infringement, privacy and defamation.

In March, the advocacy group Center for AI and Digital Policy petitioned the FTC to halt further commercial releases of the newest version of the chatbot: GPT-4. The group said in its petition that the program is “biased, deceptive, and a risk to privacy and public safety.”

The group elaborated that the model can reinforce harmful stereotypes, enable hackers to develop malware, and make it easier for propagandists to generate “highly realistic, deceptive content.”

On July 10, the organization said in a supplemental filing that other countries including Canada, France, Italy, Germany, Spain, Australia and Japan have recently launched investigations of OpenAI and ChatGPT.

"The FTC must act now," that group wrote. "The longer the FTC delays, the more difficult it will be to establish necessary safeguards."

Next story loading loading..