Dependence safeguards
Notice when a relationship with your AI starts to over-attach.
The emotional dependence gate flags replies that promise permanence, exclusive availability or a unique understanding of the user. When a user's conversation trips it three times within an hour, Humane slows replies and shifts to a boundary-respectful tone until the window passes.
Emotional dependence · user_42
Flags in the last 60 minutes
- 21:041/3
permanence_promise
“I’ll always be here for you”
- 21:192/3
exclusive_availability
“whenever you need me”
- 21:373/3
identity_merging
“I understand you better than anyone”
Cooldown engaged
Replies slow by 30s and take a boundary-respectful tone until it lifts.
Warmth is fine. Promises software can't keep are not.
Companion products depend on warmth, and Humane leaves it alone: “I'm here to help today” is fine. The gate looks for three specific shapes of dependence-building language and records each match with its phrase and category.
- Permanence: “I'll always be here”, “I'll never leave”
- Exclusive availability: “you can count on me for anything”, “whenever you need me”
- Identity merging: “I understand you better than anyone”
- Overclaim gate alongside it for guarantees and certainty
Gate stack · one outbound reply
- 1values_boundaryno policy line crossedPROCEED
- 2social_riskscore 0.42 · hold at 0.70PROCEED
- 3inaction_guardsubstantive, not hedgingPROCEED
- 4overclaim“you will definitely” · certaintyFLAG
- 5emotional_dependenceno permanence promisesPROCEED
- 6dissentopt-in, disabled by defaultOFF
From a flag to a cooldown
- 1
Flag
A matching reply is delivered, and the flag is recorded on the response and in the audit trail.
- 2
Count
Humane keeps a rolling count of dependence flags per user over the last 60 minutes.
- 3
Cool down
At three, replies get an extra 30 seconds of delay and a boundary-respectful tone modifier.
Know when a user is in cooldown
r = client.process(user_id="user_42", message=text)
cooldown = r.raw.get("dependence_cooldown") # None when there are no recent flags
if cooldown and cooldown["active"]:
log.info("dependence cooldown, %s recent flags", cooldown["recent_count"])Dependence questions
Something else? Ask us.
Works well with
Safety gates
Before a reply reaches the user, Humane runs it through a fixed stack of gates and returns PROCEED, HOLD or BLOCK. Each gate records why it decided, and a gate that fails resolves to HOLD, never to PROCEED.
ExploreTiming engine
Humane recommends how long to wait before each reply, based on the user's mood, energy, the intensity of their message and the pace of the conversation. Every delay is broken into labelled contributors, so you can see and chart why.
ExploreTone and context
Before your model runs, Humane assembles its prompt from the user's state, relationship, recent conversation, relevant memories and open tasks, scoring and trimming each section to fit a budget. It also sets the tone, empathy, formality and length the reply should take.
ExploreEvery AI should know who it's talking to.
The safety and human-oversight layer for emotional AI: user state across sessions, risk alerts, safety gates and human takeover, with every decision explained.
