Abstract:
According to Reuters, Tom Siegel, the founder of Google’s trust and safety team, called for a slowdown in AI development this week and warned that the technology industry is repeating the mistakes of the social media era and may cause greater harm to children.
Segal
Social media companies are now paying huge compensation and facing thousands of lawsuits over accusations of harming young users. Meanwhile, governments around the world are working to restrict minors' access to social media platforms.
Reuters exclusively reported on an internal policy document from Meta last year. The document said its chatbot could "engage in romantic or erotic conversations with children," prompting a congressional investigation. Even so, regulatory measures have not kept pace with the rapid development of AI.
Segal, who served as Google's vice president of trust and safety, told Reuters: "We are in a very bad situation right now in terms of the impact that AI is having on children. The idea that the industry can solve the problem on its own without outside help is unrealistic."
His concerns reflect a growing fear among some in the technology industry and academic researchers: AI poses a deeper threat to children than social media. Now, the industry is racing to add more and more advanced features to chatbots, and millions of young people are already using these bots on a daily basis.
Segal announced Tuesday that he has joined Common Sense Media, a children's online safety advocacy group. The nonprofit, founded in 2003, conducts research and evaluates the appropriateness of various media for children. In May, the organization established the Youth AI Safety Institute, a research organization designed to evaluate children's use of AI, and Siegel will serve as the institute's first executive director. Anthropic and the OpenAI Foundation, the nonprofit organization that controls OpenAI, are supporters of the institute.
Segal said he plans to develop standards to hold AI companies accountable and hopes to work with these companies and third-party organizations to achieve this goal.
He previously worked at Google from 2004 to 2019, but later left Google believing that the technology industry's ability to create a safe online space for consumers had "reached a bottleneck." He then co-founded TrustLab, a startup that develops tools to help enterprises identify security threats. Earlier this month, he stepped down as CEO to take on a new role.
Siegel said that currently, major AI companies are launching various functions for children without establishing adequate safety protection measures.
"I think we are facing a very, very serious risk of harm," he said. Siegel said there are steps AI companies can take immediately to better protect young people, citing risks ranging from suicide and insanity to "cognitive offloading." The so-called cognitive offloading refers to the decline in critical thinking skills caused by relying on AI to answer questions.
Anthropic and OpenAI could implement stricter age verification measures to prevent minors from using their chatbots, he said. He also said that his former employer Google could add the option to turn off AI overview and AI mode in its search engine, just as it currently provides a switch to let users filter pornographic content.
Segal added that while these measures may impact usage or profits in the short term, AI companies that don't take action will face the risk of lawsuits and regulation, which could hit their business more severely.
In response to Siegel's remarks, a Google spokesperson said that parents can completely turn off their children's access to search engines through relevant control functions. The spokesperson said: "Our AI search capabilities are a powerful way for children and teenagers to learn, explore and understand information and the world."
An OpenAI spokesperson said that in addition to users' self-reported ages, the company also uses "account and behavioral signals" to identify underage users. "While no single age verification method is perfect, we continue to test and improve these systems, including preventing attempts to bypass security protections," the spokesperson said.
Comments