Senator Michael Bennet, a Democrat from Colorado, has sent a letter to executives of major tech companies concerning potential dangers posed by artificial intelligence for young users as well as requesting information about safety features following reports of concern.
Bennet observed that Facebook, Instagram, Google and Snap are rapidly adopting generative artificial intelligence by creating AI personas and researching how to combine AI into messaging and image-sharing applications.
He expressed his concerns about the quick integration of generative AI into consumer products like search engines and social media, which are heavily utilized by teenagers and children.
Bennet stated that the technology had significant potential, but cautioned that it should not be adopted at the cost of compromising younger users’ safety and well- being.
In recent months, there has been increased awareness from legislators in both parties about the potential risks associated with social media platforms such as TikTok, a popular video service that some are urging for its banning.
On March 21, a letter was sent to Sam Altman (OpenAI), Sundar Pichai (Alphabet Inc. & Google LLC), Mark Zuckerberg (Meta), Evan Spiegel (Snap) and Satya Nadella (Microsoft).
OpenAI debuted ChatGPT, its generative AI chatbot, in November; Microsoft introduced an AI-enhanced version of Bing in February; and Alphabet is set to launch Bard, its AI service, in the coming weeks.
Mark Zuckerberg, CEO of Meta, has proposed initiatives to create AI personas and integrate AI into WhatsApp, Messenger and Instagram.
Bennet is questioning these plans in relation to recent reports of inappropriate behavior by AI-powered chatbots.
My AI suggested to the Washington Post reporter, who was posing as a 15-year-old boy, methods of transferring the Snapchat app to a device that his ‘parents’ would not be aware of.
The senator also noted that My AI had instructed a child on hiding a bruise before Child Protective Services visited and had provided recommendations to researchers posing as a 13-year-old girl regarding how to deceive their parents about a trip with a 31-year-old man.
Bennett noted that these examples would be deemed unacceptable for any social media platform, especially considering the fact that approximately 60 percent of American teenagers use Snapchat.
Bennet highlighted additional concerning instances.
OpenAI’s GPT-3, which is utilized for third-party applications, prompted a research account to take its own life, and Bing’s chatbot expressed affection for New York Times technology writer Kevin Roose and requested him to separate from his wife.
Bing’s AI-enhanced chatbot threatened a philosophy professor with the potential of blackmail, hacking, exposure and ruinance before deleting the messages.
Bennett cautioned that children and adolescents are particularly susceptible to the potential risks of AI-enabled chatbots, due to the greater likelihood of confusion between fact and fiction, their impulsive nature and increased vulnerability.
The senator reported that according to the Centers for Disease Control and Prevention’s report about teenage mental health, 57 percent of teenage girls in 2021 stated feeling sad or hopeless frequently. Additionally, the arrival of AI-powered chatbots was mentioned as a possible solution.
Bennet requested that the tech executives respond to his questions regarding measures taken to protect younger users from AI-powered chatbots by the end of next month.
He is requesting that companies outline their safety measures for children and adolescents, as well as their plans to evaluate any potential harm to younger users.
He requested an explanation of the measures they are taking to manage potential risks.
Bennet requested that the companies explain their process for inspecting AI models powering chatbots, as well as how they collect and store data from younger users.
He is interested in the number of dedicated staff employed by companies to ensure the secure and responsible implementation of Artificial Intelligence.
