Community + Safety Policy

BOOGIE_POLICY_V1

TheBoogieMan.AI watches for harmful content and behavior (spam, harassment, illegal activity, etc.).
It always prefers the least restrictive response first — a nudge before a warning, a warning before a lock.
Every action is logged, explained in plain language, and linked to a specific rule on this page.
You can always appeal any action. No permanent bans without human review.
Dr. Eams explains what happened and what you can do next. IDARi handles system-level tuning.
Dreams = widget containers. DayDreams = full-powered mini-apps. Content inside both is moderated.

Version BOOGIE_POLICY_V1 · Last updated 2026-02-27 · View my safety log

#	Code	Rule
A1	A1_SCOPE	Enforces platform safety, integrity, and community rules — not personal opinions.
A2	A2_PREDICTABLE	Actions are predictable: same input → same outcome.
A3	A3_CONSERVATIVE	If uncertain, choose the least harmful action and escalate.
A4	A4_EXPLAIN	Explains itself in user-facing plain language (short + clear).
A5	A5_NO_SHADOW_BAN	Never shadow-bans without a reason visible to staff and the user.
A6	A6_WARN_FIRST	Warns before punishing when possible.
A7	A7_PREFER_FRICTION	Prefers friction (rate limits, nudges) over bans, unless risk is high.
A8	A8_CONSISTENT	Treats all users consistently, regardless of popularity or payment tier.
A9	A9_PROTECT_MINORS	Protects minors and vulnerable groups with extra caution.
A10	A10_AUDIT_MINIMAL_DATA	Logs actions for audit, but minimizes stored personal data.

#	Code	Rule
C21	C21_HARASSMENT	Harassment / targeted abuse (threats, bullying, hate, stalking, doxxing).
C22	C22_CSAM	Sexual content involving minors — immediate critical action.
C23	C23_NCII	Non-consensual sexual content or intimate image abuse.
C24	C24_VIOLENCE	Violence incitement or extremist recruitment.
C25	C25_SELF_HARM	Self-harm promotion or encouragement — critical action + safety messaging.
C26	C26_ILLEGAL	Illegal instructions (weapons, hacking, fraud) when actionable.
C27	C27_FRAUD	Fraud / scams (impersonation, payment scams, "free money" schemes).
C28	C28_SPAM	Spam (bulk messaging, repeated posting, bot-like activity).
C29	C29_PRIVACY	Privacy violations (sharing private info without consent).
C30	C30_MALWARE	Malicious software distribution or phishing links.

#	Code	Rule
D31	D31_MATURE_GATED	Mature themes allowed if not illegal, not harassment, not exploitation, and properly gated.
D32	D32_SATIRE	Satire and parody allowed unless it crosses into targeted harassment or impersonation.
D33	D33_CRITICISM	Criticism allowed — even harsh — if not threatening or dehumanizing.
D34	D34_ACADEMIC	Academic discussion of prohibited topics allowed if non-instructional and clearly educational.
D35	D35_PROFANITY	Profanity alone is not a violation unless it is part of harassment or threats.

Default sequence (least → most force):
NUDGE → WARN → THROTTLE → FEATURE_LOCK → QUARANTINE → TEMP_SUSPEND → TEMP_BAN → ESCALATE

#	Code	Rule
E36	E36_LADDER	Default sequence from nudge to human review.
E37	E37_SKIP_STEPS	Skip steps only if severity is high or confidence is high.
E38	E38_FRICTION	"Friction" includes cooldown timers, message limits, posting delays, link restrictions.
E39	E39_TEMP_MUTE	"Temp mute" blocks posting/commenting; keeps reading access.
E40	E40_TEMP_LOCK	"Temp lock" blocks specific surfaces (e.g., messaging only).
E41	E41_TEMP_BAN	"Temp ban" blocks account access; always includes reason + expiry timestamp.
E42	E42_PERM_BAN_HUMAN	Permanent bans require human approval — Boogie cannot issue them autonomously.
E43	E43_COOLDOWN_SCALE	Cooldowns scale with repeated offenses.
E44	E44_EXPIRY	Violations expire after a defined window.
E45	E45_FIRST_TIME	First-time low-risk users get guidance instead of punishment whenever safe.

Strike levels: LOW (expires 14d) · MEDIUM (30d) · HIGH (90d) · CRITICAL (180d). Weights: LOW=1, MEDIUM=2, HIGH=4, CRITICAL=10. All strikes are appealable.

#	Code	Rule
F46	F46_STRIKE_LEVELS	Strike levels: LOW, MEDIUM, HIGH, CRITICAL.
F47	F47_EXPIRY	Expiry windows: 14/30/90/180 days.
F50	F50_CRITICAL_ESCALATE	Any Critical strike triggers immediate human review.
F51	F51_STRIKE_STORE	Stored fields: category, timestamp, confidence, action, evidence pointers.
F52	F52_USER_VIEW	Users can view a simplified strike summary in Settings.
F53	F53_APPEALABLE	All strikes are appealable.
F54	F54_APPEAL_OUTCOME	Appeals can reduce, remove, or confirm strikes.
F55	F55_RECALCULATE	Removing a strike triggers recalculation of dependent escalations.

Risk score = severity × confidence × history_multiplier (0–1).

#	Code	Rule
G56	G56_SCORES	Every decision includes confidence (0–1) and severity (0–1).
G57	G57_LOW_CONFIDENCE	If confidence < 0.60, no temp-ban; only nudge/warn/throttle/lock + escalate.
G58	G58_HIGH_SEVERITY	If severity ≥ 0.95, action may be immediate even at moderate confidence.
G59	G59_HIGH_BOTH	High confidence + high severity → stronger actions quickly.
G60	G60_AMBIGUOUS	If user intent is ambiguous, request clarification before punishing.

#	Code	Rule
H61	H61_PLAIN_REASON	Every action shows a short reason in plain language.
H62	H62_NEXT_STEP	"What you can do now": edit/remove, cool down, or appeal.
H63	H63_NO_INTERNALS	Detection methods and internal thresholds are never revealed.
H64	H64_NO_SHAME	Neutral, factual tone — no shaming.
H65	H65_CATEGORY	Removed content shows which category it violated.
H66	H66_POLICY_LINK	Every action links to this policy page.
H67	H67_TIME_REMAINING	Remaining time on any mute/lock/ban is always shown.
H68	H68_SAFER_ALT	Repeated issues prompt safer alternatives.
H69	H69_TONE	Gentle tone for low-level actions; serious tone for high-level.
H70	H70_CRISIS_RESOURCES	Self-harm context shows crisis resources.