Openai launches a safety guidance system, parental controls on Chatgpt

Openai started a new test Safety guidance system In Chatgpt during the weekend, on Monday, parental controls were presented to Chatbot – drawing mixed reactions from users.

Safety features come in response to several accidents to some ChatGPT models that check the imaginary thinking of users instead of redirect harmful conversations. It faces Openai a Illegal death lawsuit One liked one, after the death of a teenager’s suicide, after months of interactions with ChatGPT.

The guidance system is designed to discover sensitive conversations emotionally and automatically switch to GPT-5-ahinking, which the company considers the best model equipped for high-risk safety work. In particular, GPT-5 models were trained with A new safety feature called Openai “safe completion”, “ Which allows them to answer sensitive questions in a safe way, rather than just refusing to participate.

It is in contrast to the company’s previous chat models, which are designed to be acceptable and answer questions quickly. GPT-4O has been subjected to special scrutiny due to its excessive nature, which fed all the accidents caused by artificial intelligence and drawing a large base of loyal users. When Openai GPT-5 as the default in AugustPaying many users back and demanded to reach the GPT-4O.

Although many experts and users welcomed safety features, others have criticized what they see as a very cautious application, as some users accuse Openai of adult treatment such as children in a way that degrades the quality of service. Openai suggested that obtaining it will take time and give itself a 120 -day period of repetition and improvement.

Nick Torley, Vice President and head of the ChatGPT application, acknowledged some “strong reactions to 4O responses” due to the implementation of the router with interpretations.

“The guidance occurs on the basis of each means; the switching of the default model occurs on a temporary basis,” Torley post on X. “ChatGPT will tell you which model is active when asking it. This is part of a wider effort to enhance guarantees and learn from using the real world before starting wider row.”

Received the implementation of parent control tools in ChatGPT Similar levels of praise and contemptWith the praise of some giving parents a way to maintain the tabs on the use of their children from artificial intelligence, while others are afraid that the door to Openai will open adults like children.

Controls allow parents to allocate teenage experience by setting quiet hours, turning off the audio and memory, removing images, and hacking for models. Teenage accounts will also get Additional content protection -Such as low graphics content and the ideals of extreme beauty-and the detection system that realizes possible signs that a teenager may consider self-harm.

“If our systems discover a possible damage, a small team of people trained specifically reviews the situation,” according to Openai’s. Blog. “If there are signs of sharp distress, we will contact the two men through the email and the text message and pay the alert on their phones, unless they choose.”

Openai acknowledged that the system will not be perfect and may sometimes raise warnings when there is no real danger, “but we believe it is better to act and alert a parent so that they can intervene instead of silence.” Artificial Intelligence Company said it is also working on ways to access law enforcement services or emergency services if she discovers an imminent threat to life and one of the parents can not be accessed.