Free and open first
Best AI Security Testing Tools in 2026
AI security testing tools let you probe AI systems for weaknesses before those systems are put to work. Compare custom test cases with data leakage, jailbreak, prompt injection, and unsafe output tests to see which checks align with your concerns. Deployment mode, free plan availability, and paid-from pricing offer additional factors for weighing fit. OpenSecureAI Scanner, Project Moonshot, and Promptfoo are among the named options, alongside PromptGuard and PyRIT. Consider the risks you want to assess and whether each tool’s listed deployment and plan details suit your testing needs.
- 29 apps
- Research read 23 Sept 2026
- How the order is worked out
- 1
AIRTA Red Team
Free plan Open source APILinuxMac+3AIRTA Red Team offers broad runtime assessment workflows and report evidence without a license fee. Its deprecated, unmaintained status and required API key are important constraints.
- 2
CyberSecEval
Free plan Open source APILinuxMac+2CyberSecEval covers several cybersecurity risks as well as defensive capabilities, with distinct benchmarks for code, prompt injection, patch generation, and SOC reasoning.
- 3
Giskard
Free plan Open source APILinuxSelf-hosted+1 - 4
OpenSecureAI Scanner
Free plan Open source APILinuxMac+3OpenSecureAI Scanner offers several ways to check untrusted text, with additional hosted detections for authenticated scans.
- 5
Promptfoo
Free plan Open source APILinuxMac+3Promptfoo combines prompt evaluation with red-team testing and a broad provider and integration list.
- 6
Augustus
Free plan LinuxMacSelf-hosted+1Augustus provides a broad set of model vulnerability probes, detectors, and report formats in a self-hosted package.
- 7
AgentScan
Free plan APIWebAgentScan provides API-based security scans with report exports and a free monthly allowance. Note the stated OWASP coverage omissions and authorization requirement when planning assessments.
- 8
Basilisk
Free plan LinuxMacWindowsBasilisk offers broad testing, provider, reporting, and CI options, with local processing for key workflows. Users must obtain written authorization before testing systems they do not own.
- 9
garak
Free plan LinuxMacSelf-hosted+1garak provides configurable, automated probing for language-model weaknesses across a range of integrations. Its text-only assessment scope is important for anyone evaluating broader model outputs.
- 10
AgentDojo
Free plan APILinuxSelf-hostedAgentDojo provides components for building and evaluating LLM agent security benchmarks, including custom attacks and defenses.
- 11
AIRTaaS
Free plan WebAIRTaaS combines recurring AI security checks with risk reports and progression tracking.
- 12
Crucible
Free plan LinuxSelf-hosted - 13
Project Moonshot
Free plan APILinuxMac+3Project Moonshot provides multiple ways to benchmark and red-team LLMs, including custom evaluations and machine-readable results.
- 14
DeepTeam
Free plan APISelf-hostedChoose DeepTeam for open-source LLM red teaming with broad vulnerability coverage and integrations for development pipelines. Its dynamic attacks and standards support structured security checks.
- 15
PromptGuard
Free plan APIExtensionLinux+4PromptGuard combines LLM request screening with deployment choices from managed cloud to air-gapped operation.
- 16
PyRIT
Free plan APILinuxMac+3PyRIT provides a modular way to assess generative AI systems with repeatable scenarios and a range of targets.
- 17
HarmBench
Free plan APILinuxSelf-hostedHarmBench provides a configurable pipeline for measuring attacks, defenses, and red-teaming performance.
- 18
LLMMap
Free plan LinuxChoose LLMMap if you need a free, self-hosted Linux tool for authorized prompt injection testing with multiple prompt families and backends.
- 19
RedForge
Free plan APILinuxSelf-hostedChoose RedForge if you need a self-hosted, open-source LLM security scanner with broad provider support, compliance reporting, and CI-oriented SARIF output.
- 20
RAMPART
Free plan APILinuxMac+1Choose RAMPART if you want a pytest-native way to test agent safety, including cross-prompt injection and expected behavior.
- 21
PromptRAID
Free plan LinuxWindowsPromptRAID is a useful fit for teams or researchers who need a self-hosted command-line framework for prompt-injection red teaming and reports of test outcomes.
- 22
InjectLab
Free planInjectLab provides structured adversarial tests, scoring ideas, and mitigation guidance for LLM security work.
- 23
Red Team LLM Lab
Free plan Self-hostedWindowsRed Team LLM Lab offers a focused, self-hosted toolkit for adversarial LLM checks and reporting. Its Windows requirements and planned roadmap features matter when evaluating deployment scope.
- 24
SuperRed
Free plan LinuxSuperRed provides modular, self-hosted testing for chatbot and agent security.
- 25
Akto Agentic Red Teaming
Paid APISelf-hostedWebChoose Akto Agentic Red Teaming if your team needs continuous adversarial testing across AI agents and applications, with customizable probes and broad connector coverage.
#1Places #1, #2, #3, #4, #5 on this list can be bought by the makers of the software on it. See the price