Thoughtox is a private space for honest, unfiltered thought processing. We built it on the premise that people need somewhere to write what they can't say out loud — without being judged, scored, or exposed. This policy exists to protect that space for everyone who uses it.
The rules here are narrow on purpose. We are not trying to police your internal state. We are trying to prevent the small category of use that makes the space unsafe or unworkable for others.
1. What Thoughtox Is For
Private reflection. Writing through something difficult. Processing an event before you're ready to talk to anyone about it. Anything that fits that description is permitted — regardless of how dark, messy, or unresolved the content is.
Thoughtox is a single-user tool. Your entries are private to you by default. There is no public feed, no community forum, and no social layer. The rules below exist for the small number of cases where that private use bleeds into harm.
2. Prohibited Use
The following are not permitted under any circumstances:
- Content targeting others for harm. Direct threats, harassment, or content designed to intimidate a specific person or group.
- Child sexual abuse material (CSAM). Any content that sexualises minors. Reports are made to relevant authorities without exception.
- Hate speech. Content that dehumanises individuals or groups based on race, religion, ethnicity, gender, sexuality, disability, or national origin.
- Solicitation of illegal acts. Using Thoughtox to plan, coordinate, or solicit criminal activity.
- Automated abuse. Scripted submission, scraping, API stress testing, or any attempt to extract training data from the AI pipeline.
- Security circumvention. Attempting to bypass authentication, rate limits, moderation filters, or encryption. This includes prompt injection attacks against the AI processing pipeline.
- Multiple accounts to evade restrictions. Creating additional accounts after suspension, or to bypass free-tier limits.
Note what is not on this list: dark thoughts, self-criticism, expressions of rage, grief, suicidal ideation, or any other difficult internal state. Writing those things is exactly what Thoughtox is for. The prohibited list targets actions and outputs that harm others — not the emotional content of private reflection.
3. How Moderation Works
Every submission passes through an automated pipeline before it reaches the AI reflection engine. Here is exactly what that pipeline does:
PII Redaction
Phone numbers, email addresses, and similar personal identifiers are automatically detected and redacted from the stored text. This protects you — if you write a phone number or address in a thought, it won't sit in our database in plaintext. The redaction is noted on the entry but the original value is not retained.
Crisis Detection
Content containing language associated with immediate self-harm or suicidal crisis is flagged. When flagged, AI analysis is skipped and you are shown local support resources instead. The entry is saved. The crisis flag is stored as private metadata on your account — it is not reported to any external authority, employer, insurer, or anyone else. It exists only to surface help to you in that moment.
We want to be explicit: writing about suicidal thoughts, past self-harm, or dark ideation is not a violation of this policy. The crisis detection is a safety net, not a punitive system.
Content Classification
An intent classifier screens for the prohibited categories listed in Section 2. Entries classified as potentially violating are either quarantined (blocked from AI processing, stored but not surfaced to you) or flagged for limited review depending on confidence level. You will see a message if your entry is blocked.
The classifier is imperfect. If you believe an entry was incorrectly blocked, contact toxthotox@gmail.com with the approximate time of submission.
What Moderation Does Not Do
- It does not run general manual reading across everyone's entries.
- Authorized staff may review a small number of escalated entries for abuse, safety, or AI-output reports.
- It does not share flagged entries with third parties unless legally compelled.
- It does not use your content to train classifiers or AI models.
- It does not judge your emotional state or restrict access based on mood.
4. Enforcement
Violations of this policy may result in a warning, temporary suspension, or permanent account termination depending on severity. For most first-time violations we will warn before acting. For CSAM, direct threats, or coordinated abuse, termination is immediate and without notice.
If you believe your account was restricted in error, contact us at toxthotox@gmail.com. We will review it.
Questions about this policy? toxthotox@gmail.com