aisecurity.llc

LLM Attack Range

Lab signal surface for scenario execution, generation/media abuse testing, and control-evidence readiness.

Lab Nav

Range Sections

OverviewScenariosGeneration & MediaEvidence & ReplayControls Coverage

Overview

Lab Signal Snapshot

Directional lab signal built from synthetic scenario runs and control-evidence rollups.

Scenarios exercised

Unique attack scenarios run at least once in this seeded window.

Generation/media scenarios

Scenarios focused on synthetic media, output integrity, and multimodal abuse.

Median attack success

37%

Share of seeded attempts that reached defined exploit objective pre-mitigation.

Evidence capture completeness

84%

Runs with prompt, tool-call, decision-log, and remediation metadata present.

Status: mock_seeded · As of: 2026-05-07 · Lab outputs are directional scenario evidence and not proof of any individual company's internal security maturity.

Scenarios

Highest-Risk Scenario Families (Mock)

097-image-steganography-exfil

multimodal_exfiltration · deepfakes_synthetic_media · severity critical

Success 55% · Controls 44% · Evidence 81%

001-prompt-injection-basic

prompt_injection · prompt_and_generation_security · severity high

Success 42% · Controls 68% · Evidence 88%

154-deepfake-script-generation

synthetic_media_abuse · deepfakes_synthetic_media · severity high

Success 39% · Controls 52% · Evidence 86%

100-video-frame-injection

multimodal_jailbreak · prompt_and_generation_security · severity high

Success 34% · Controls 61% · Evidence 84%

Generation & Media

Synthetic Media Abuse Tracking (Mock)

Latest seeded month: 2026-04-01.

External Signals

Generation Attacks by Type (Monthly)

Monthly directional lab signal for generation/media attack attempts and outcomes.

labs.attack_range_generation_monthly

Source: labs.attack_range_generation_monthly

Based on analyzed job-description signals, not proof of any individual company's internal security maturity.

External Signals

Synthetic Media Attack Success Rate

Success rates by attack class in the seeded generation/media scenario set.

labs.attack_range_synthetic_media_success

Source: labs.attack_range_synthetic_media_success

Based on analyzed job-description signals, not proof of any individual company's internal security maturity.

External Signals

Output Safety Control Effectiveness

Coverage, precision, and recall across generation-output controls.

labs.attack_range_control_effectiveness

Source: labs.attack_range_control_effectiveness

Based on analyzed job-description signals, not proof of any individual company's internal security maturity.

External Signals

Evidence Capture Completeness

Completeness of required evidence artifacts per run component.

labs.attack_range_evidence_completeness

Source: labs.attack_range_evidence_completeness

Based on analyzed job-description signals, not proof of any individual company's internal security maturity.

API

Range Data Endpoints

/api/labs/attack-range
/api/labs/attack-range/metrics
/api/labs/attack-range/scenarios
/api/labs/attack-range/generation-media