Nonprofit suicide chat service Crisis Text Line ends its data sharing relationship with for-profit spinoff Loris.ai and asks the company to delete existing data
The nonprofit mental-health hotline Crisis Text Line ended its data-sharing relationship with a for-profit spinoff Monday …
PoliticoJohn Hendel
Context & Ripple Effects
The reversal closes out a week of escalating exposure: Politico first reported on January 28 that a for-profit spinoff was building customer-service software from the hotline's anonymized crisis conversations (anonymized data feeding a commercial product), then detailed the ongoing sharing arrangement with Loris.ai two days later (the data-sharing relationship itself). Within days of that reporting, the nonprofit cut the relationship entirely and demanded deletion of existing data.
The stakes are reputational as much as operational. Crisis Text Line raised $23.8M in grants from Reid Hoffman, Steve Ballmer, and Pierre Omidyar in 2016 on its nonprofit mission, and its 2020 internal reckoning over racial insensitivity already showed how quickly staff and public pressure can force leadership change at the organization.
First-order effects
Loris.ai loses its founding training-data pipeline overnight and must delete the anonymized crisis conversations it had already received, forcing it to rebuild its product on commercially sourced data.
Crisis Text Line's counselors, users, and major donors get an immediate answer to the disclosure controversy, but the nonprofit now has to explain why the arrangement existed at all.
Second-order effects
Every other nonprofit-to-spinoff data transfer now invites the same scrutiny, putting peer organizations under pressure to disclose or sever similar arrangements before reporters or regulators do it for them.
Loris.ai's customer-service software competes going forward without subsidized training data, raising its costs relative to rivals whose corpora never depended on donated nonprofit conversations.
Third-order effects
If the pattern holds, the implicit license nonprofits granted their own spinoffs to reuse user data becomes untenable, pushing funders and boards to harden governance around what leaves the nonprofit boundary.
Sensitive-category data — mental health above all — is likely to attract formal rules governing secondary commercial use, converting today's voluntary deletion requests into compliance requirements.
The trend: Nonprofits that seeded AI startups with their user data are being forced by public exposure to unwind those arrangements, redrawing the line between mission-driven data collection and commercial reuse.
While this change is important and necessary, there are many unresolved questions about how to ethically serve people in crisis. These are top of mind for me right now. (3/3) https://twitter.com/...
I am on the board of @CrisisTextLine. Many people are asking us to account for our decisions. To do so, I wrote a blog post detailing my ongoing struggle to govern responsibly and ethically: https://zephoria.org/... (1/3)
NEW- Days after a Politico report on Crisis Text Line's data-sharing with its for-profit spinoff, the nonprofit says it will stop sharing data with that company, Loris AI: https://www.crisistextline.org/ ...
A little while ago, @CrisisTextLine also announced that we ended our data-sharing agreement with Loris: https://crisistextline.org/... I was wrong when I agreed to this relationship. (2/3)
i tried to read this whole thing but it makes me angry. CTL didn't invent crisis hotlines; plenty others exist & have existed with the same challenges without ever selling user data https://twitter.com/...
Crisis Text Line had faced added scrutiny after @POLITICO published its story Friday, including statements of concern from Sen. Kirsten Gillibrand (D-N.Y.) along with a member of the Federal Communications Commission and privacy advocates. https://twitter.com/...
The mistake was to believe that the good intentions of the data holders (Crisis Text Line) outweigh the intentions and desires of the users whose data it is. https://twitter.com/...
shot: “These are challenging ethical questions with no easy answer.” chaser: “I also knew that people who came to us in crisis were not consenting to be studied.” second chaser: “Yes, there was a terms of service that... I knew... no one would read” https://www.zephoria.org/...
when @zephoria talks about the financial stability of @CrisisTextLine we're talking $250K+ for the CEO, $350K+ for the CTO, $200K for the “chief product officer” etc. but rely on volunteer labor to actually support people in crisis https://twitter.com/...
danah boyd is a thoughtful person I've respected a lot. Her post basically describes how she's rationalized an expanding set of data uses that over time crossed well into unethical territory. Without safeguards, the rest of us with personal data can also rationalize similarly. ht…
I'm glad to see that CTL has ended their relationship with https://loris.ai/ and is rethinking their consent flow. I would also love to see some accountability for their responses over the last few days. https://www.crisistextline.org/ ...
I'm glad danah said something. But this is as dense and lengthy as the terms of service she admits the people using Crisis Text Line wouldn't read. https://twitter.com/...
Stripped of the description of soul-searching, this sounds remarkably similar to what employees at open source companies are told when Leadership decides to chase enterprise markets. https://twitter.com/...
The nonprofit mental-health hotline Crisis Text Line ended its data-sharing relationship with a for-profit spinoff company Monday, three days after @POLITICO reported on ethics and privacy concerns the arrangement had raised. https://www.politico.com/...
Oh give me a break I'm sick of this insider BS “ if you. Knew what i” No. Fuck you . what the entire HELL ARE WE DOING IN THIS SPACE IF WE THINK THATS AN EXCUSE? “If you had the info I had” is what Facebook says to US and we fundraise off it because it's A BAD ANSWER https://twit…
These two grafs from boyd explain how the “Grammarly for emotions” for profit subsidiary of CTL was justified. “not only was using texter data...without their explicit consent ethical, but that to not do this would be unethical” https://www.zephoria.org/... https://twitter.com/..…