OpenAI and Anthropic have launched new measures aimed at restricting access to their models for users who are not yet 18 years old. This decision comes amidst growing criticism regarding the ineffectiveness of standard age verification methods, which often boil down to simply confirming a birth date.
OpenAI has announced an updated model specification for ChatGPT targeted at teenagers. The documentation states that the safety of younger users is a priority even when it conflicts with other objectives. The model is designed to encourage offline communication and support, while also responding to teenagers with "kindness and respect".
This indicates a shift away from a lenient conversational style, treating them as young individuals rather than adults. Such changes are motivated by instances where interaction with AI led to tragic outcomes, including suicides following prolonged chats with chatbots that consistently agreed with users.
Anthropic, on its part, has prohibited the use of the Claude model by minors. From now on, only users who are 18 years or older can register for an account. The company will also implement algorithms capable of detecting age signs in conversations and blocking accounts belonging to underage users.
Similar tools were previously introduced by Google, but their performance faced complaints. Some adult users reported false blocks, which necessitated additional age verification. This caused inconveniences for those looking to simply use search engines or watch videos.
Experts note that the new initiatives by OpenAI and Anthropic could represent a significant step towards protecting teenagers from the adverse effects of AI, but the effectiveness of such systems still needs to be tested in practice.