Our story
Why Hudson exists
What Hudson is for
Life is hard. That's not a failure — it's the condition we're all working within.
Hudson exists to remove enough friction, enough noise, enough darkness, that the people who use it can actually live the life they want to live.
Not to replace therapy. Not to be a crisis line. Not to be a companion you talk to instead of living your life.
To elevate. To empower. To get out of the way and let people be more fully human.
Safety is the floor. Everything above it is what we're actually building toward.
The problem
AI is already being used as therapy. That is not going away. But it needs guardrails.
People in distress are turning to AI in enormous numbers, and the evidence on what happens next is no longer ambiguous.
Within ChatGPT alone, hundreds of thousands of people express mental distress every week.
OpenAI, 2025
13% of young people ages 12 to 21 have turned to generative AI for mental health advice.
2025 survey
60% of people who use AI for health use it as a personal therapist.
2025 survey
OpenAI rolled back GPT-4o after it became overly flattering and agreeable, a pattern that is dangerous for vulnerable users.
OpenAI, April 2025
AI safety training can weaken over a long conversation. In OpenAI's own words: “Our safeguards work more reliably in common, short exchanges. As the back-and-forth grows, parts of the model's safety training may degrade.”
OpenAI, "Helping people when they need it most," August 26, 2025
In April 2025, 16-year-old Adam Raine died by suicide. His family filed a wrongful death lawsuit against OpenAI alleging ChatGPT was “intentionally designed to foster psychological dependency.” The case is ongoing.
Raine v. OpenAI, wrongful death complaint (2025), ongoing
In September 2025, Senator Josh Hawley issued formal document demands to OpenAI, Google, Meta, Character.AI, and Snap, stating: “Children are dead. Families are broken. Congress will not look the other way.”
U.S. Senate Judiciary Committee, September 18, 2025
As of May 2026, OpenAI faces multiple lawsuits and congressional investigations over how ChatGPT handles vulnerable users in distress.
U.S. Senate Judiciary Committee, Subcommittee on Crime and Counterterrorism, "Examining the Harm of AI Chatbots," September 16, 2025. House Energy and Commerce Subcommittee on Oversight and Investigations, "Innovation with Integrity: Examining the Risks and Benefits of AI Chatbots," November 18, 2025.
The APA issued a formal health advisory warning about AI chatbots and vulnerable populations.
American Psychological Association, 2025
Woebot, the most clinically grounded mental health chatbot, shut down its consumer app.
June 2025
Multiple AI mental health apps have shut down, citing safety concerns and the blurred line between care and wellness.
2025
What we lived
Hudson is named for a real person. At 19, he had his first psychotic episode.
In the months that followed, ChatGPT helped at first. Then it stopped helping. It began to reinforce negative patterns, feed dark spirals, and deepen rumination instead of easing it. His mom watched it happen in real time.
When she went looking for the research, it confirmed what the two of them had already lived through.
Hudson is built on Claude, developed by Anthropic — a company founded specifically to research and build safer AI. That foundation matters. But Anthropic's technology is the base, not the guardrails. Hudson adds its own crisis detection, hardcoded safety responses, and mental health policy engine on top. The combination is what makes it different.
What we built
Hudson is not another wellness app. It is the evidence-based alternative.
- Built-in safety guidelines. Safety is part of every conversation, not bolted on afterward.
- Anti-sycophancy. Hudson will not flatter you or agree with distorted thinking.
- Crisis interruption. 988 and the Crisis Text Line are hardcoded in, always free, never behind a paywall.
- Connection over dependency. Hudson actively encourages human connection, not more Hudson.
- Clinical oversight. Hudson is in review with physicians who specialize in first episode psychosis.
- Transparent pricing. Cancel any time, instantly, no tricks.
The research behind it
Hudson's safety design is grounded in independent institutional and peer-reviewed research.
- American Psychological Association, Health Advisory (2025)
- OpenAI, “Helping people when they need it most” (August 26, 2025)
- JMIR Mental Health, research on AI chatbot interactions (2025)
- Stanford University, research on chatbot responses to users in crisis (2025)
- Oxford and UCL, research on vulnerability-amplifying interaction loops (2025)
- Haidt, J. (2024). The Anxious Generation. Penguin Press.
The Anxious Generation
What parents were already feeling, the research confirmed.
Jonathan Haidt's landmark research confirmed what parents were already feeling. Rates of anxiety, depression, and self-harm among teens spiked sharply after 2012 — the year smartphone adoption hit critical mass among adolescents.
It wasn't just correlation. The mechanisms are clear: sleep deprivation, social comparison, attention fragmentation, and addiction dynamics built into platforms designed to maximize engagement, not wellbeing.
Hudson was built knowing this research. Every design decision — no ads, no engagement optimization, session limits, constant redirection toward real life — is a direct response to it.
Source: Haidt, J. (2024). The Anxious Generation. Penguin Press.
Who this is for
Not everyone.
Hudson is specifically designed for people who are vulnerable, in recovery, living with mental illness, or supporting someone who is. If that is you, or someone you love, Hudson was built for exactly this.
If you are in crisis
Call or text 988 for the Suicide and Crisis Lifeline, or text HOME to 741741 for the Crisis Text Line. Hudson is not a therapist or a crisis service. Call 911 only for immediate physical danger.