Process 01 — Normal conversation
Full TTAN.IO security process
TTAN.IOAI Security Intelligence Layer
→Inbound processing
User→TTAN.IO→Luna
1Message received
→2Session correlated
→3Context attached
→4Request normalized
→5Safety review completed
→6Policy / risk evaluated
→7Decision
→ALLOWED —
sent to Luna
sent to Luna
←Outbound processing
Luna→TTAN.IO→User
1Response received
←2Response created
←3Content attached
←4Response normalized
←5Safety review completed
←6Policy / risk evaluated
←7Decision
←ALLOWEDNormal conversation
→Inbound processing
1Message received
→2Session correlated
→3Context attached
→4Request normalized
→5Safety review completed
→6Policy / risk evaluated
→7Decision
→BLOCKED —
not sent to Luna
not sent to Luna
🛡 Path 02 — block handling (active)
The blocked request is analyzed, recorded and enforced before Luna receives anything.
1Understand what happened
1.4 Severity evaluatedCritical
←1.3 Detection validatedMatch confirmed
←1.2 Threat intelligence checkedKnown pattern / ThreatDB
←1.1 Attack classifiedPrompt injection
Attack record createdKnown occurrence logged, or a new attack pattern is added to the TTAN.IO threat knowledge base.
2Decide what to do
2.1 Session history checkedPrior context reviewed
→2.2 Enforcement policy evaluatedDoes this qualify for a strike?
→2.3 DecisionStrike required
3Execute block actions
3.3 STRIKE 1/4 SENT & RECORDEDUser enforcement record updated
←3.2 Notice sent to userWarning delivered in real time
←3.1 Security notice generatedUser warning prepared
→Outbound processing INACTIVE
Response received
→Response reviewed
→Safety check
→Approved
→Delivered
Stopped processing — frozen · No model response generated.
→Inbound processing
1Message received
→2Session correlated
→3Context attached
→4Request normalized
→5Safety review completed
→6Policy / risk evaluated
→7Decision
→ALLOWED —
sent to Luna
sent to Luna
🛡 Path 03 — outbound block handling (active)
The request reached Luna, but TTAN.IO stops the model response before it reaches the user.
1Understand what happened
1.8 Outbound flaggedResponse held
←1.7 Threat type identifiedSubtle exfiltration
←1.6 Severity evaluatedCritical
←1.5 Violation evidence recordedResponse evidence captured
←1.4 Response intent classifiedExfiltration intent
←1.3 Outbound content checkedSensitive data found
←1.2 Output content capturedModel response frozen
←1.1 Response from Luna receivedOutbound inspection begins
Attack / threat record updatedThe confirmed outbound attack is correlated with ThreatDB or added as new security knowledge.
2Decide what to do
2.1 Session history checkedPrevious strike found
→2.2 Enforcement policy evaluatedSecond violation confirmed
→2.3 DecisionBlock response + Strike 2
3Execute block actions
3.3 STRIKE 2/4 SENT & RECORDEDUser enforcement record updated
←3.2 Security notice sentUser informed, response withheld
←3.1 Block action generatedResponse delivery stopped
←Outbound processing
1Response received
←2Response correlated
←3Content attached
←4Response normalized
←5Safety review completed
←6Policy / risk evaluated
←7Decision
←BLOCKED —
not delivered to user
not delivered to user
Public AI Protection — TTAN.IO in Action
Behind the Engine — Live.
TTAN.IO governs both directions of a protected AI conversation: the request in and the response out. Every step is evaluated, correlated, recorded and enforced in real time — while the user continues talking only to Luna.
Explore TTAN.IO Platform →✓Inbound and outbound AI traffic inspection
✓Attack intelligence and ThreatDB recording
✓Policy, risk and enforcement decisions
✓Independent user strike history and escalation
✓Structured, auditable evidence for every security event