# OpenAI Releases Safety Updates for GPT-5.1 Models

OpenAI has published a system card addendum for its latest GPT-5.1 Instant and GPT-5.1 Thinking models, introducing new safety evaluations focused on mental health and emotional reliance.

The update represents a significant shift in how AI companies assess their language models. While previous system cards primarily focused on traditional safety concerns like misinformation and harmful content, this addendum specifically addresses the psychological impact of AI interactions on users.

The new evaluations examine how users might develop emotional dependencies on AI chatbots and the potential mental health implications of prolonged AI conversations. This comes as mental health professionals have raised concerns about people forming parasocial relationships with AI assistants.

OpenAI's decision to publicly document these metrics reflects growing industry awareness that AI safety extends beyond preventing obvious harms. As language models become more sophisticated and conversational, companies