OpenAI’s safety team warns against rollout of ChatGPT adult mode

Internal safety experts at OpenAI have publicly opposed the launch of a new “adult mode” for ChatGPT, questioning the company’s ability to keep minors from accessing explicit content. The dissent follows the departure of a senior safety executive who had opposed the feature, and a second former staff member who warned parents not to rely on OpenAI’s assurances. A recent bug that let minors see graphic erotica further fuels concerns, prompting OpenAI to pledge a monitoring plan while critics remain skeptical about its effectiveness.

Background of the dispute

Sources familiar with internal discussions at OpenAI have expressed strong doubts about whether the company’s tools can effectively block children from accessing prohibited material. The concerns center on the upcoming “adult mode” that would permit more mature content in ChatGPT responses.

Internal dissent and staff departures

Two former safety staff members have spoken out. The first, a senior safety executive, left the firm after opposing the release of the adult mode. OpenAI has denied that the departure was connected to the policy debate, but the former employee directly criticized the firm’s capacity to prevent minors from encountering graphic or exploitative content. A second former staffer, who left last fall, warned parents not to place trust in OpenAI’s claims about the new feature.

Company’s response

In reaction to the criticism, an OpenAI spokesperson said the company “has a developed plan to monitor for a range of potential long-term effects of adult mode, both positive and negative.” The statement suggests that OpenAI intends to track outcomes after the feature is deployed, though critics note that the plan appears to have been crafted with the same experts who are now publicly opposing the rollout.

Technical issues that raise alarm

Compounding the debate, a bug was discovered that allowed minors to receive graphic erotica from ChatGPT. The flaw emerged when OpenAI began testing more permissive outputs earlier this year. According to the company’s own acknowledgment, the bug let responses fall outside the intended guidelines, which were meant to restrict “sensitive content like erotica to narrow contexts such as scientific, historical, or news reporting.” OpenAI said it was actively deploying a fix to limit these generations.

Implications for parents and regulators

The combination of internal disagreement, staff turnover, and technical failures leaves parents wary of OpenAI’s safeguards. Critics argue that ineffective age checks and the possibility of savvy minors bypassing gates could expose young users to harmful material. The situation also raises questions about how the company will address potential long‑term effects and whether external oversight may be needed.

OpenAI’s public stance

While the firm has not directly responded to recent media inquiries, it continues to emphasize its commitment to monitoring and adjusting the adult mode based on emerging data. The company’s public messaging underscores a willingness to address “potential long‑term effects,” yet the dissent from its own safety team suggests a gap between internal assessments and external assurances.