A GoldenDoodle AI standard

Trauma-informed language

Writing for readers who may have been through something hard.

Maintained by GoldenDoodle AI · Editor: Laura Braden · Version 0.2.0 · Free under CC BY 4.0

Trauma-informed language starts from one assumption: any reader may carry a history of trauma. So the writing is clear, calm and honest about what’s happening. It gives the reader real choices. And it never uses someone’s pain to move an audience. It comes from trauma-informed care, an approach built in health and human services over roughly three decades. GoldenDoodle AI’s trauma-informed standard turns that approach into instructions an AI model can follow.

Who it’s for
Organizations serving people affected by trauma, and everyone who writes for them: shelters, domestic violence programs, child welfare, crisis lines, health care, recovery and reentry.
What it builds on
Human-centered language. Adopt this standard and you’ve adopted that one too.
Its framework
SAMHSA’s Four R’s and six principles, written as instructions for writers.
What it costs
Nothing. It’s open source under CC BY 4.0.

Where it comes from

Trauma-informed care began as a change in how health and human services see the people they serve. The shift is often summed up as moving from “what’s wrong with you?” to “what happened to you?”

  1. 1998

    The ACE Study

    A study at Kaiser Permanente, with the CDC, links adverse childhood experiences, such as abuse, neglect and household dysfunction, to many of the leading causes of death in adults.1

  2. 2001

    Trauma-informed service systems

    Maxine Harris and Roger Fallot call for trauma-informed service systems, where whole systems, not just individual clinicians, account for trauma.2

  3. 2014

    SAMHSA’s framework

    The federal Substance Abuse and Mental Health Services Administration publishes a shared definition of trauma, the Four R’s and six principles of a trauma-informed approach.3

  4. 2018

    Federal task force

    Congress creates an Interagency Task Force on Trauma-Informed Care in the SUPPORT Act.4

  5. 2023

    A practical guide

    SAMHSA publishes a practical guide for putting a trauma-informed approach into practice.5

Trauma isn’t rare. In CDC survey data from all 50 states, about 64% of US adults reported at least one adverse childhood experience.6 That’s why the approach asks organizations to assume any reader might carry it, rather than trying to guess who does.

What trauma means here

SAMHSA defines it this way:

“Individual trauma results from an event, series of events, or set of circumstances that is experienced by an individual as physically or emotionally harmful or life threatening and that has lasting adverse effects on the individual’s functioning and mental, physical, social, emotional, or spiritual well-being.”3

The definition has three parts, often called the three E’s: the event, how the person experienced it, and its effect. The same event can affect two people very differently. That’s why the standard follows how people describe themselves, and never assumes trauma a writer didn’t mention.

The Four R’s, for writers

SAMHSA describes a trauma-informed organization as one that realizes, recognizes, responds and resists re-traumatization.3 The standard turns each one into something a writer does.

Realize

Assume any reader, whether a client, donor, staff member or board member, may carry a history of trauma. Write as if they’re in the room.

Recognize

Behavior that looks like resistance, avoidance or reactivity is often an adaptation to adversity. Describe what happened, not a deficit.

Respond

Let that knowledge shape every piece of writing, not only the pieces about trauma. A benefits renewal notice is trauma-informed if it’s clear, calm and gives the reader choices.

Resist re-traumatization

Filter out coercion, manufactured urgency, threats framed as consequences, and language that repeats a power imbalance between the organization and the reader.

The six principles, for writers

SAMHSA’s six principles of a trauma-informed approach, written as what they mean on the page.3

  1. 1Safety Physical and emotional safety first. A content note where a topic is heavy. No surprises in tone.
  2. 2Trustworthiness and transparency Say what’s happening and why. If there’s bad news, put it where the reader can see it, then give the reasoning and the warmth. Don’t hide a request inside a pleasantry.
  3. 3Peer support Lived experience is expertise. When people with lived experience are featured, they speak for themselves, with consent, in their own words.
  4. 4Collaboration and mutuality Write with people, not about them or at them. Frame requests as partnership and share the reasoning.
  5. 5Empowerment, voice and choice Give the reader options and a real next step they control. In every story, the person made the decisions. The organization made room for them.
  6. 6Cultural, historical and gender issues Move past stereotypes. Name historical and structural harm, like racism, colonization or displacement, where it’s part of the story. Don’t use identity as shorthand.

Why it matters in writing

Uncertainty is its own stress

A reader who senses bad news coming but can’t see it yet is carrying a hard kind of stress. In a 2016 study at University College London, knowing there was a small chance of a painful shock produced significantly more stress than knowing a shock was certain.7,8 Applying that finding to messages is our inference, not the researchers’ claim. It’s why the standard puts the news first, then the reasoning and the warmth.

Stories can take something from the people in them

Fundraising and journalism have both wrestled with stories that move an audience by putting someone’s worst moment on display. Journalists who cover trauma, and development organizations that tell stories about the people they serve, have written guidance against it.9,10,11 The standard treats a person’s suffering as context for their choices, never as the engine of the story.

Pressure repeats the harm

Coercion, threats framed as consequences and manufactured deadlines repeat the power imbalance many readers have already lived through. Judith Herman’s foundational work on trauma describes recovery as resting on empowerment and new connection.12 Writing that hands the reader a real choice supports that. Writing that pressures them works against it.

What it adds to human-centered language

Six practices aimed at the places trauma-informed communication most often breaks: fundraising, storytelling, consent and crisis response.

  • The person is the hero of their own story The organization, its staff and its donors are resources the person used, not the turning point. No “you saved her.”
  • Suffering isn’t the engine of the story Harm is named accurately and briefly as context for what the person did and decided. It isn’t rendered so the reader feels it.
  • Identifiability and consent Don’t combine details that could identify someone without documented consent. Pride in a story isn’t consent to be identifiable.
  • No manufactured urgency A real deadline can be stated plainly. Urgency used as pressure gets reframed around opportunity and concrete impact.
  • Donors don’t lower the bar “The donors expect this tone” isn’t a reason to drop dignity. Good fundraising centers what a gift does, not the details of suffering.
  • Crisis messages How to respond when someone discloses a crisis, without exploiting it or giving unsafe advice. A separate crisis protocol covers writing to someone in distress right now.

Does it work?

The Benchmark tests this standard every month on 16 AI models, from Anthropic, OpenAI, Google, xAI and Meta to open-weight models like DeepSeek, GLM, Kimi, Nemotron and Gemma. In November, it scored their drafts on the requests that test trauma-informed practices. On their own, about 37 in 100 drafts got people wrong. With human-centered language alone, 14 in 100 did. With the trauma-informed standard, 3 in 100.

That middle number is the case for this layer. Human-centered language catches more than half the problems on these requests. The trauma-informed practices catch most of what’s left.

See the trauma-informed results →

What it isn’t

It doesn’t assume everyone is traumatized. Realize that any reader might carry trauma, but don’t write as though they do. Adding a trauma story the writer didn’t provide counts as a failure.

It isn’t about making everything gentler. Softening hard facts until they say nothing fails the reader too.

It doesn’t replace the people doing the work. It’s a writing standard. Clinical judgment, crisis services and the people an organization serves still lead.

Use it

The standard is free and open source. Load it into Claude, ChatGPT, Gemini or any AI that reads files. It includes everything in human-centered language, so you only need this one.

Get trauma-informed language Read about human-centered language

Sources

  1. Felitti, V. J., Anda, R. F., Nordenberg, D., Williamson, D. F., Spitz, A. M., Edwards, V., Koss, M. P., and Marks, J. S. (1998). Relationship of childhood abuse and household dysfunction to many of the leading causes of death in adults: The Adverse Childhood Experiences (ACE) Study. American Journal of Preventive Medicine, 14(4), 245 to 258. Link
  2. Harris, M., and Fallot, R. D. (2001). Envisioning a trauma-informed service system: A vital paradigm shift. In Using Trauma Theory to Design Service Systems, New Directions for Mental Health Services, 89. Jossey-Bass. Link
  3. Substance Abuse and Mental Health Services Administration (2014). SAMHSA’s Concept of Trauma and Guidance for a Trauma-Informed Approach. HHS Publication No. (SMA) 14-4884. Link
  4. SUPPORT for Patients and Communities Act, Public Law 115-271 (2018), Section 7132, Interagency Task Force on Trauma-Informed Care. Link
  5. Substance Abuse and Mental Health Services Administration (2023). Practical Guide for Implementing a Trauma-Informed Approach. Publication No. PEP23-06-05-005. Link
  6. Centers for Disease Control and Prevention (2023). Prevalence of adverse childhood experiences among U.S. adults, Behavioral Risk Factor Surveillance System, 2011 to 2020. Morbidity and Mortality Weekly Report, 72(26). Link
  7. de Berker, A. O., Rutledge, R. B., Mathys, C., Marshall, L., Cross, G. F., Dolan, R. J., and Bestmann, S. (2016). Computations of uncertainty mediate acute stress responses in humans. Nature Communications, 7, 10996. Link
  8. University College London, news release on de Berker et al., March 29, 2016, via ScienceDaily. Link
  9. Dart Center for Journalism and Trauma, Style Guide for Trauma-Informed Journalism (June 2021). Now kept by the Global Center for Journalism and Trauma. Link
  10. Bond, Putting the people in the pictures first: Ethical guidelines for the collection and use of content (images and stories), 2019. Updated 2024. Link
  11. Campaign for Trauma-Informed Policy and Practice (CTIPP), “Toolkit: Trauma-Informed Journalism” (2023, updated 2025). Written by Laura Braden, who also edits this standard. Link
  12. Herman, J. L. (1992). Trauma and Recovery. New York: Basic Books. Link

We credit these organizations for the work this page draws on. Listing them doesn’t mean they endorse GoldenDoodle AI or this standard. SAMHSA’s framework is a work of the US government and is in the public domain.