A former Google trust and safety executive has warned that the artificial intelligence industry risks repeating mistakes made during the rise of social media, with AI potentially posing greater risks to children than social media did.
Tom Siegel, who previously served as a vice president of trust and safety at Google, told Reuters that AI companies could not be expected to address risks to young users without stronger external oversight.
Siegel has joined Common Sense Media, a US-based nonprofit that researches and evaluates media for children. He will serve as the first executive director of its Youth AI Safety Institute, which was launched in May to study children’s use of artificial intelligence.
He said the risks associated with AI could include severe mental and behavioural harms, as well as reduced critical thinking when young people increasingly rely on AI systems to provide answers.
Siegel said companies including OpenAI and Anthropic could strengthen age verification systems to limit access by underage users. He also suggested that Google could give users clearer options to disable AI-generated features in its search products.
His comments come as technology companies continue to expand AI-powered products used by children and teenagers, while governments and regulators debate how existing rules should apply to the technology.
Siegel previously worked at Google from 2004 to 2019 before co-founding TrustLab, a company that developed tools for detecting online safety risks. He stepped down as TrustLab’s chief executive earlier this month to take up his new role at Common Sense Media.
The Youth AI Safety Institute is backed by organisations including Anthropic and the OpenAI Foundation.
Siegel said one of his priorities would be developing independent standards for testing the safety of AI products, similar to systems used to assess the safety of cars.
Google said parents can already restrict or disable children’s access to its search engine, while an OpenAI spokesperson said the company uses account and behavioural signals alongside self-reported age to identify underage users.
OpenAI said no single age-assurance method is perfect and that it continues to improve its systems, including measures designed to prevent users from bypassing safeguards.
Anthropic did not immediately respond to a request for comment.
The debate comes as AI companies accelerate the development of increasingly capable models and agents, raising questions about how quickly safety standards and regulation can adapt to the technology.
