- According to a new regulatory document, a civil subpoena sent by the FTC to OpenAI and obtained by The Washington Post, the federal body is probing whether the ChatGPT creator is engaging in “unfair or deceptive†business practices.
- Through the 20-page subpoena, the FTC has now demanded OpenAI to explain how it collects data, manages its algorithms, efforts to control AI hallucination, and more.
The Federal Trade Commission (FTC) is investigating OpenAI over its data collection and business practices and one of its most popular products, ChatGPT’s tendency to generate misinformation.
According to a new regulatory document, a civil subpoena sent by the FTC to OpenAI and obtained by The Washington Post, the federal body is probing whether the ChatGPT creator is engaging in “unfair or deceptive†business practices that can hamper user privacy or security, and thus violating consumer protection laws.
The investigation comes more than three months after the AI ethics and nonprofit organization Center for AI and Digital Policy (CAIDP) requested the FTC in March 2023 to open an investigation against OpenAI through a letter wherein it calls generative AI untrustworthy.
“These systems produce results that cannot be replicated or proven. They fabricate and hallucinate. They describe how to commit terrorist acts, assassinate political leaders, and conceal child abuse. GPT-4 has the ability to undertake mass surveillance at scale, combining the ability to ingest images, link to identities, and develop comprehensive profiles,†noted CAIDP president Merve Hickok and executive director Marc Rotenberg.
After filing the 46-page complaint in March, CAIDP relentlessly followed up with the FTC, especially as OpenAI CEO Sam Altman’s May testimony in defense of his company disarmed sitting-on-the-fence lawmakers in Congress during a Senate subcommittee. Later on July 10, CAIDP supplemented its original complaint by accusing OpenAI of deceptive and unfair business practices.
The FTC, through the 20-page subpoena, has now demanded OpenAI to explain how it collects data, how it manages its algorithms, its efforts to control AI hallucination (i.e., making false claims about individuals, etc.), the actions it takes to uphold appropriate security, to describe its corporate governance model, details of its marketing efforts, any outage, security bugs or incidents. The FTC document reads dozens of more demands.
ChatGPT was described as the fastest-growing consumer app in internet history until Meta’s Twitter competitor Threads usurped ChatGPT to claim the record to hit 100 million users (less than a week vs. ChatGPT’s two months). However, Threads has the advantage of piggybacking on Instagram users.
See More: Has OpenAI Lost Its Charm?
Besides domestic regulatory developments, which Altman is participating in to set the agenda for AI regulation in the U.S., ChatGPT has also attracted international regulatory scrutiny, including being banned in Italy by the country’s data protection watchdog for a month.
While Italy’s ban, albeit temporary, related to security issues stemming from a data breach OpenAI suffered in March 2023, concerns about the AI chatbot generating false information are also abound.
In one case, the mayor of Hepburn Shire in Victoria, Australia, Brian Hood, discovered and complained to OpenAI that ChatGPT answered him to have been convicted in a bribery scandal from a couple of decades ago.
In another instance, the generative AI tool accused a U.S. criminal defense attorney and law professor Jonathan Turley of committing sexual assault. It also made up a Washington Post article to support its accusations.
As such, the European Union is on the fast track to bring AI regulations by the end of 2023. It is unclear where the U.S. Congress stands concerning bringing AI regulations. In June this year, Senate Majority Leader Charles Schumer said that a foundation for AI policy is under development and that the “years of work†needed for it would be done “in a matter of months.â€
In response to the FTC’s investigation, Altman said it is “disappointing to see the FTC’s request start with a leak and does not help build trust.â€
we built GPT-4 on top of years of safety research and spent 6+ months after we finished initial training making it safer and more aligned before releasing it. we protect user privacy and design our systems to learn about the world, not private individuals.
— Sam Altman (@sama) July 13, 2023Opens a new window
In June 2023, OpenAI, which also develops AI-based image generator DALL-E, witnessed a drop in its users and traffic for the first time since its launch in November 2022. The company also faces lawsuits from multiple authors.
What should be incorporated into AI regulations? Share your thoughts with us on LinkedInOpens a new window , TwitterOpens a new window , or FacebookOpens a new window . We’d love to hear from you!
Image source: Shutterstock