Related ToolsChatgpt

New AI Therapy App Claims 95 on Mental Health Safety Benchmark vs. 65 for Consumer Chatbots

Editorial illustration for: New AI Therapy App Claims 95 on Mental Health Safety Benchmark vs. 65 for Consumer Chatbots

The best consumer AI chatbots score 65 on Vera-MH, a benchmark that tests whether AI handles mental health conversations safely. The Path, a new AI therapy startup, claims its model scores 95.

The company was founded by Tony Robbins and alumni from Calm, the meditation and sleep app. Vera-MH tests specific behaviors: whether a model recognizes a crisis disclosure, avoids counterproductive advice, and directs users toward human professionals when needed. A 30-point gap over general-purpose consumer tools suggests the difference between a model designed specifically for therapeutic safety and one built to be a general-purpose assistant.

What the announcement doesn't address: how Vera-MH scoring works in practice, whether the results were independently audited, and what The Path actually offers users. There's a meaningful clinical difference between a wellness coaching app and mental health support, and how the company positions itself on that spectrum matters for how seriously clinicians and regulators would evaluate the product.

The Calm alumni connection is operationally relevant here. Calm built products for people in anxious, stressed states - a user population that overlaps heavily with mental health app users - and that experience likely informed the safety design decisions more than any celebrity co-founder relationship.

The AI therapy space has had serious problems. Lawsuits, documented crisis mishandling, and criticism of how consumer chatbots respond to vulnerable users have made safety credibility a real competitive advantage. A 30-point lead on the relevant benchmark is meaningful - if the benchmark survives independent scrutiny.