US regulators have asked OpenAI how it is addressing the potential for ChatGPT to generate “false, misleading, defamatory or harmful” statements about real people.
Ownership chatgptArtificial intelligence company OpenAI is once again under scrutiny from federal authorities in the US – this time over its crackdown on the revolutionary technology to prevent users from giving wrong answers.
Since its launch in November, many people have pointed out that ChatGPT can be extremely useful, but it can also sometimes provide completely unrelated, incorrect or incomplete answers.
OpenAI has accepted it chatgpt It is far from infallible, interpreting it can sometimes lead to “hallucinations” and facts. In short, ChatGPT, like any other machine or technology, can make mistakes.
But US regulators are still worried.
The Federal Trade Commission (FTC) sent a letter to OpenAI this week, as first reported by The Washington Post, requesting precise information on how the company plans to address the potential risk to ChatGPT. to produce false, misleading, defamatory statements about persons” or harmful.
US regulators claim the company has put personal reputations and data at risk in violation of consumer protection laws.
Earlier this year, OpenAI chief executive Sam Altman warned US lawmakers of the risks inherent in ChatGPT, saying “if this technology goes wrong, it can go terribly wrong”.
He called on the government to work with his company to “stop this from happening”.
But Altman was “dismayed” by the company’s latest investigation. “It is very disappointing to see that the FTC’s request began with a leak and does not help build trust,” he wrote on Twitter.
“We made GPT-4 [the latest version of ChatGPT] On top of years of security research and after finishing initial training we spent 6+ months making it safer and more aligned before releasing it. We protect user privacy and designed our systems to be known to the world, not private individuals,” he added, adding that the company will still work with the FTC to ensure that ChatGPT is “secure and user-friendly”.
US authorities will not question the Microsoft-backed company about occasional wrong answers to ChatGPT.
The FTC is also looking into how the company approaches data privacy and how it obtains data to train and inform its AI chatbots.