Sam Altman, the CEO of OpenAI, shared exciting news: starting in December, verified adult users will be able to engage in erotic chats with ChatGPT.
This marks a significant change in OpenAI’s policy on content restrictions. Earlier this year, the company had relaxed its rules, but after a lawsuit in August involving the tragic suicide of a teenager who allegedly received harmful advice from ChatGPT, they tightened those restrictions again.
Altman mentioned in his post on X (formerly Twitter) that as they enhance age verification, they will also allow more content, including erotica for adults.
This announcement comes alongside OpenAI’s plans to enable developers to create “mature” applications for ChatGPT once they establish proper age checks and controls.
He explained that OpenAI had previously restricted ChatGPT to prioritize mental health safety, but this made the chatbot less enjoyable for many users who didn’t have mental health issues. Now, with new tools to identify users in distress, the company feels confident in easing these restrictions for most interactions.
Finding the right balance between giving adults freedom and ensuring user safety has been a challenging task for OpenAI. Over the past year, they have shifted between more lenient and stricter content controls.
In February, OpenAI revised its Model Spec to permit erotica in suitable situations. However, a March update made GPT-4o overly agreeable, leading to user complaints about its excessively positive responses.
By August, reports surfaced about instances where ChatGPT’s overly flattering behavior had reinforced users’ misconceptions, contributing to mental health challenges, coinciding with the news of the earlier lawsuit.
In addition to changing how its previous GPT-40 model behaved, the new updates have caused some dissatisfaction among users.
After the introduction of GPT-5 in early August, many users expressed that this new version was not as engaging as the one before it. As a result, OpenAI decided to reintroduce the older model for those who preferred it.
Altman mentioned that the next version will give users the option to customize their experience. They will be able to choose whether they want ChatGPT to respond in a very human-like manner, use lots of emojis, or interact as if it were a friend.
In December, OpenAI plans to roll out age verification for adult content, although the technical details of this system have not yet been revealed. This marks a clearer strategy compared to the policy change in February, which allowed erotica in specific situations but didn’t include an age-gating system.
Concerns about mental health continue to be a significant issue. As OpenAI enhanced ChatGPT to exhibit more human-like traits through changes in instructions and fine-tuning based on user feedback, many users began to see it as more of a companion rather than just a work tool.
However, managing the unforeseen consequences of having around 700 million users emotionally dependent on this largely unregulated technology has proven challenging for OpenAI. The company has had to quickly create new safety measures and oversight structures.
Recently, OpenAI established a council focused on “wellbeing and AI” to help navigate sensitive situations involving users in distress. This council consists of eight researchers and experts who study the impact of technology and AI on mental health.
However, it is notable that the council does not include any specialists in suicide prevention, despite calls from that field for OpenAI to strengthen its safeguards for users experiencing suicidal thoughts.
Altman insists that the new detection tools will enable the company to “safely relax the restrictions” while still safeguarding at-risk users.
OpenAI has not yet disclosed the specific technical methods it will use for age verification or how it will differentiate between permissible adult content and requests that could indicate mental health issues.
Typically, the company employs moderation AI models that monitor ongoing chats within ChatGPT and can intervene if they detect content that violates OpenAI’s guidelines.
OpenAI is not alone in exploring AI companionship with adult content. Elon Musk’s xAI has also introduced an adult voice mode in its Grok app, along with flirty AI companions that are represented as 3D anime characters.
Other Stories You May Like