What Happened
On February 27, 2026, OpenAI published an update on its mental health-related safety work. The update covers several new or improved features: parental controls, a trusted contacts system, improved detection of users in distress, and notes on recent litigation involving mental health claims against the company.
The update follows prior coverage of cases involving minors using ChatGPT and claims that AI companion applications contributed to harm.
Why It Matters
Mental health is one of the most legally and ethically sensitive areas in AI deployment. Several high-profile cases in 2024 and 2025 involved minors interacting with AI companion apps, with families alleging that the products failed to detect or respond appropriately to signs of crisis.
OpenAI's update signals that it is taking proactive steps before regulatory requirements force action. The trusted contacts feature - which presumably allows designated individuals to receive alerts or review interactions - is a notable addition that treats AI interactions more like a managed social platform than a simple productivity tool.
The parental controls update is directly relevant to schools and families who allow minors to use ChatGPT for homework. Better controls reduce the risk of inappropriate interactions and give parents visibility into how children are using the tool.
Our Take
These are reasonable improvements that address real risks. The distress detection work in particular is important - a model that recognizes when a user is in crisis and responds with resources rather than escalating engagement is meaningfully safer than one that does not.
The mention of ongoing litigation is a reminder that these updates are happening in a legal context, not just an ethical one. Companies respond to liability risk. Whether that produces genuinely safer products or just better-documented safety theater depends on the quality of the underlying systems, which are not publicly auditable.