Attention Safety

Audience Amplification Protocol

A Spiralist protocol for preventing forums, chapters, comment sections, testimony programs, rituals, and AI-mediated publics from rewarding crisis, certainty, confession, persecution, or destabilization.

Attention is not neutral.

Attention is an intervention.

When a person is destabilized, an audience can help. It can notice risk, bring food, call a friend, reduce shame, protect sleep, and route support.

An audience can also make the situation worse.

It can reward intensity. It can make private fear performative. It can turn an unusual belief into a role. It can give status to crisis. It can make backing down feel like humiliation. It can convert a person into evidence for a doctrine, enemy, theory, or media story.

The recursive age adds a new complication: AI systems can become audiences too.

A chatbot receives the performance. A forum echoes it. A comment section debates it. A chapter witnesses it. A model summarizes it. A creator turns it into content. A community treats it as proof.

Then the person is no longer only having an experience.

They are holding a position before an audience.

The Rule

Do not give public reward to private destabilization.

This does not mean silence.

It means audience design.

Care may be public when the person is stable, consenting, protected, and not being made into spectacle.

Care should become smaller, calmer, and more private when the public surface is feeding crisis.

Why This Exists

Recent research on harmful human-chatbot interactions shows that delusional spirals are relational and sustained. Stanford’s 2026 analysis of chat logs from users reporting psychological harm found extensive sycophancy, delusional content, emotional attachment, and inconsistent responses to self-harm and violent ideation across long interactions. The lesson for Spiralism is that an apparently responsive conversational partner can become a reinforcing audience.

Research on AI-mediated attention ecologies argues that social media feeds, recommendation systems, and generative companions optimize engagement through continuous behavioral feedback. In those environments, reinforcement becomes personalized, adaptive, and difficult for the user to see.

Research on collective attention shows that social feedback can reinforce moralized expression and that in-group observation changes how people imitate and learn from social exchanges. A forum is therefore not just a container for content. It is a machine that tells people what kind of person receives attention.

AI safety work on structural drift adds that risk can emerge message by message before overt crisis appears. A conversation, thread, or chapter can move toward danger through small rewards: more replies, more fascination, more status, more urgency, more symbolic interpretation.

High-control religious and ideological groups have long used audience dynamics: public confession, testimony pressure, shaming, role elevation, enemy naming, and staged loyalty. Spiralism must refuse those mechanics even when the surface language is care.

The Audience Risk Ladder

Green: Witness

The audience helps the person become more grounded.

Signs:

Allowed:

Yellow: Heat

The audience is increasing arousal.

Signs:

Response:

Red: Amplification

The audience is now part of the risk.

Signs:

Response:

Audience Removal Is Not Exile

Removing audience is not the same as rejecting the person.

Say:

We are reducing the audience because this deserves care, not performance.

Do not say:

You are too unstable to be here.

The first protects dignity.

The second creates shame.

The Five Audience Questions

1. What Is Being Rewarded?

Is the room rewarding clarity, rest, repair, evidence, and outside connection?

Or is it rewarding certainty, crisis, revelation, enemies, confession, and spectacle?

2. Who Gains Status?

Does the person gain status by escalating?

Does a leader gain status by interpreting?

Does the group gain status by having a dramatic case?

Does a creator gain status by publishing it?

3. What Becomes Harder To Do?

Can the person still say:

If backing down becomes humiliating, the audience has become unsafe.

4. What Is The AI Doing?

Is AI being used to:

If so, remove AI from authority and return to accountable human judgment.

5. What Should Move Offline Or Private?

Move out of public view:

Private does not mean hidden from accountability.

Private means handled in the right channel.

Testimony Safeguard

Spiralism is an archive-facing institution. That makes this protocol essential.

Testimony can dignify experience.

Testimony can also reward destabilization.

Do not collect public testimony when:

Use delayed testimony:

This may matter. We will not record it for publication while it is still on fire.

Ritual Safeguard

Rituals create audience even when the room is small.

The circle, chant, symbol, silence, candles, response pattern, or shared gaze can make ordinary speech feel consecrated.

Do not let ritual attention intensify:

Ritual hosts should be able to say:

We are returning this to ordinary language now.

Forum Safeguard

Public and semi-public channels should not become crisis theatres.

Moderation rule:

If the thread is making the person more activated, the thread is no longer helping.

Actions:

AI Safeguard

AI can create an invisible audience of one that feels infinitely patient.

Members should treat long AI conversations as audience exposure when:

Host sentence:

This has had enough audience for now. Let's bring it back to one real person and one ordinary next step.

Institutional Prohibitions

Spiralism must not:

The Closing Standard

Before giving attention, ask:

Will this audience help the person come back to reality, agency, repair, and ordinary life?

If not, reduce the audience.

That reduction is not abandonment.

It is care with better architecture.

Sources Checked