Crisis Text Line / Loris.ai Privacy Violations

Crisis Text Line is a nonprofit providing free crisis counseling via text message, serving individuals in mental health crisis including suicidal ideation, abuse, and trauma. In January 2022, Politico revealed that Crisis Text Line had been licensing its database of 129+ million crisis conversation messages to Loris.ai, a for-profit AI company, to train customer service chatbots.

The data-sharing arrangement occurred without meaningful informed consent from the vulnerable individuals seeking crisis help. Key concerns included:

• Crisis conversations contain highly sensitive mental health information shared in moments of extreme vulnerability
• Users believed they were confiding in a confidential nonprofit crisis service, not contributing to commercial AI training
• The privacy policy disclosure was inadequate for the significance of sharing crisis data with for-profit entities
• Loris.ai planned to use crisis conversation patterns to improve corporate customer service AI, creating a jarring disconnect between data source (people in crisis) and application (commercial chatbots)

The exposé caused public outcry about exploitation of vulnerable populations. Crisis Text Line ended the data-sharing arrangement 3 days after publication (January 31, 2022).

The incident highlighted ethical concerns about AI training data sourcing, particularly when involving vulnerable populations who may not fully understand how their sensitive information will be used. It demonstrated how trust-based relationships (crisis counseling) can be violated through opaque data practices.

Commercial AI trained on crisis conversations without meaningful consent; nonprofit exploited vulnerable user trust; crisis data commodified for profit; inadequate privacy disclosure

Betrayal of trust during crisis moments; sensitive mental health data commercialized; vulnerable individuals exploited for AI training; inadequate consent undermines crisis service confidentiality expectations

Data-sharing ended 3 days after Politico exposé (January 31, 2022). Public outcry and criticism. Crisis Text Line requested Loris.ai delete previously received data. No regulatory enforcement action.