ChatGPT It is necessary to be able to actively hang up dangerous conversations, and repeat offenders will have their accounts immediately banned. In the past, it was up to the users themselves to stop, but now AI grabs the microphone first.

ChatGPT It is necessary to proactively hang up dangerous conversations, and repeat offenders will have their accounts immediately banned. In the past, it was up to the user to stop the conversation, but now AI snatches the microphone first. OpenAI recently announced that in the next few weeks, it will add stronger violent content gates to ChatGPT: when the conversation continues to revolve around the idea of ​​violence or serious harm to others, the model can end the conversation on its own; if the same account is blocked multiple times due to such issues, the account will be banned. This is not a rule from scratch - ChatGPT originally prohibited providing guidance on planning or committing violence. This upgrade has added "active cut-off" to the execution link, moving from "passive refusal to answer" to "active cut-off". The platform's processing logic for dangerous content has obviously moved forward. All age groups are treated equally. The new regulations extend the restrictions to all users and do not allow access to the Internet based on age. Adjust the response scale that focuses on "historical attacks" - the model will reduce or even refuse to provide details from a security perspective, covering the weapons, equipment and acquisition methods used in the attacks, and also limit the horizontal comparison of the scale of casualties that may amplify the visibility of the perpetrators. In other words, adults and minors face the same set of tightened boundaries. The difference is only in depth: adult users will also be blocked, but there will no longer be additional practical information blocking of weapons and security. Minors add another layer. For accounts identified as under 18 years of age, OpenAI additionally blocks practical information about weapons and security systems that can be used to commit violence. This will take effect even if no violent intent is apparent on the surface of the question. Teenagers are still allowed to use ChatGPT to ask for help—how ​​to prevent or de-escalate violence, or contact a trusted adult—are not blocked. The accompanying improved age prediction will combine multiple signals such as user behavior to continuously determine whether the account may be underage, and automatically apply youth protection if necessary; if it is misjudged, it can also actively submit identity information for verification. via AI News (author: AI Base)