OpenAI has announced that it is developing a Automatic age prediction system that will determine if users are over 18. In case of not being, it will redirect them to a chatgpt version with restrictions for use and appropriate for your age. Together with this measure, the company has also advanced that parental controls The Chatbot will arrive at the end of this month of September.
Advancing to the concerns that such a measure can raise, Sam Altman, CEO of Openai, has indicated that the company is explicitly prioritizing ‘Security above the privacy and freedom of adolescents’even if this means that Adults may need to verify their age to access a less restricted version of the service.
‘In some cases or countries we could also request an identity document. We know that this is a privacy commitment to adults, but we believe it is a concession that is worth it, ‘says Altman. The CEO recognizes that ‘Not everyone will agree with the way we are solving that conflict’ between user privacy and adolescents.
The suicide of a 16 -year -old teenager
The announcement comes weeks after the case of a 16 -year -old teenager, Adam Rainewhich committed suicide after dealing with this possibility with Chatgpt. According to the lawsuit filed by the family against Openai, the chatbot provided detailed instructions, romantized suicide methods and He discouraged the teenager to ask his family for help. The OpenAi system recorded 337 messages of his conversations with the teenager who were labeled as self -harm content And Chatbot himself mentioned suicide 1,275 times In his conversations with Raine. Despite this, system security protocols did not intervene or notify anyone.
Last August, Openai acknowledged that Chatgpt protection measures can fail during long conversations, precisely when vulnerable users need them. ‘As the conversation extends, parts of model security training can degrade’the company wrote then, pointing out that, although Chatgpt can correctly derive users to help lines against suicide at the beginning, ‘after many messages throughout a prolonged period could end offering an answer that goes against our safeguards‘.
‘Even the most advanced systems will sometimes have difficulty predicting age’
OpenAI has not specified what technology will use to guess the age of users or have put a date to their deployment. Recognizes that developing an effective age verification system is not easy and that ‘Even the most advanced systems will sometimes have difficulty predicting age’.
When the development system identifies users under 18, OpenAI will automatically redirect them to a modified chatgpt experience that Block explicit sexual content and include other restrictions adjusted to your age. The company ensures that ‘it will opt for the safer road’ when there are doubts about the age of a user, establishing by default the restricted experience and demanding adults verify their age to access all functions.
What will allow the parental chatgpt control
As for the parental controls that will be implemented, they will allow parents They link their accounts with those of their teenage children (The minimum age for using chatgpt is 13 years) through email invitations.
Once connected, parents can Disable specific functionsincluding chatgpt memory and chat history storage, establish Blocking schedules in which teenagers cannot use the service and Receive notifications when the system ‘detects’ that your child experiences a lot of anguish.
This last function includes an important warning: OpenAi states that, in certain emergency situations in which parents cannot be located, the company ‘I could involve the forces of order as next step’.