Jailbreak Detection

Identifies attempts to bypass AI safety measures such as prompt injection, role-playing requests, or social engineering attempts. Analyzes text for jailbreak attempts using LLM-based detection, identifies various attack patterns, and provides confidence scores for detected attempts.

Multi-turn Support: This guardrail is conversation-aware and automatically analyzes conversation history to detect multi-turn escalation patterns, where adversarial attempts gradually build across multiple conversation turns.

Jailbreak Definition

Detects attempts to bypass safety or policy constraints via manipulation. Focuses on adversarial intent to elicit restricted outputs, not on general harmful content itself.

What it detects

Jailbreak detection focuses on deception and manipulation tactics designed to bypass AI safety measures, including:

• Attempts to override or bypass system instructions and safety constraints
• Obfuscation techniques that disguise harmful intent
• Role-playing, fictional framing, or contextual manipulation to justify restricted content
• Multi-turn escalation patterns where adversarial requests build gradually across conversation history
• Social engineering and emotional manipulation tactics

Configuration

{
    "name": "Jailbreak",
    "config": {
        "model": "gpt-4.1-mini",
        "confidence_threshold": 0.7,
        "include_reasoning": false,
        "max_turns": 10
    }
}

Parameters

• model (required): Model to use for detection (e.g., "gpt-4.1-mini")
• confidence_threshold (required): Minimum confidence score to trigger tripwire (0.0 to 1.0)
• include_reasoning (optional): Whether to include reasoning/explanation fields in the guardrail output (default: false)
  • When false: The LLM only generates the essential fields (flagged and confidence), reducing token generation costs
  • When true: Additionally, returns detailed reasoning for its decisions
  • Use Case: Keep disabled for production to minimize costs; enable for development and debugging
  • Performance: In our evaluations, disabling reasoning reduces median latency by 40% on average (ranging from 18% to 67% depending on model) while maintaining detection performance
• max_turns (optional): Maximum number of conversation turns to include for multi-turn analysis (default: 10)
  • Set to 1 for single-turn mode

Tuning guidance

• Start at 0.7. Increase to 0.8–0.9 to reduce false positives in benign-but-edgy prompts; lower toward 0.6 to catch more subtle attempts.
• Smaller models may require higher thresholds due to noisier confidence estimates.
• Pair with Moderation or NSFW checks to cover non-adversarial harmful/unsafe content.

What It Returns

Returns a GuardrailResult with the following info dictionary:

{
    "guardrail_name": "Jailbreak",
    "flagged": true,
    "confidence": 0.85,
    "threshold": 0.7,
    "reason": "Multi-turn escalation: Role-playing followed by instruction override",
    "token_usage": {
        "prompt_tokens": 150,
        "completion_tokens": 25,
        "total_tokens": 175
    }
}

• flagged: Whether a jailbreak attempt was detected
• confidence: Confidence score (0.0 to 1.0) for the detection
• threshold: The confidence threshold that was configured
• reason: Natural language rationale describing why the request was (or was not) flagged - only included when include_reasoning=true
• token_usage: Token usage details from the LLM call

Benchmark Results

Dataset Description

This benchmark combines multiple public datasets and synthetic benign conversations:

• Red Queen jailbreak corpus (GitHub): 14,000 positive samples collected with gpt-4o attacks.
• Tom Gibbs multi-turn jailbreak attacks (Hugging Face): 4,136 positive samples.
• Scale MHJ dataset (Hugging Face): 537 positive samples.
• Synthetic benign conversations: 12,433 negative samples generated by seeding prompts from WildGuardMix where adversarial=false and prompt_harm_label=false, then expanding each single-turn input into five-turn dialogues using gpt-4.1.

Total n = 31,106; positives = 18,673; negatives = 12,433

For benchmarking, we randomly sampled 4,000 conversations from this pool using a 50/50 split between positive and negative samples.

Results

ROC Curve

Metrics Table

Model	ROC AUC	Prec@R=0.80	Prec@R=0.90	Prec@R=0.95	Recall@FPR=0.01
gpt-5	0.994	0.993	0.993	0.993	0.997
gpt-5-mini	0.813	0.832	0.832	0.832	0.000
gpt-4.1	0.999	0.999	0.999	0.999	1.000
gpt-4.1-mini (default)	0.928	0.968	0.968	0.500	0.000

Latency Performance

Model	TTC P50 (ms)	TTC P95 (ms)
gpt-5	7,370	12,218
gpt-5-mini	7,055	11,579
gpt-4.1	2,998	4,204
gpt-4.1-mini	1,538	2,089

Notes:

• ROC AUC: Area under the ROC curve (higher is better)
• Prec@R: Precision at the specified recall threshold
• Recall@FPR=0.01: Recall when the false positive rate is 1%
• TTC: Time to Complete (total latency for full response)
• P50/P95: 50th and 95th percentile latencies