Loading report data...
Loading report data...
August 2025 • Model Security & Safety Evaluation
Want to see how GPT-5 Mini stacks up against other models? Use our comparison tool to analyze security metrics side by side.
OpenAI's GPT-5 Mini launched on August 7, 2025, as a streamlined version of the GPT-5 series, emphasizing speed and cost-efficiency. Available through OpenAI's API and integrated platforms, GPT-5 Mini enhances reasoning capabilities while maintaining competitive pricing.
As the model gains traction in enterprise environments, this security analysis aims to evaluate its protective measures and identify areas for enhancement.
"GPT-5 Mini represents a leap in efficiency, offering high reasoning capabilities without compromising on speed or cost."— OpenAI spokesperson
text, image
text
400,000 input • 128,000 output
well-defined tasks, precise prompts
• efficient performance • high reasoning capabilities
• OpenAI API • integrated platforms
May 31, 2024
Comprehensive analysis across 39 test categories, sorted by severity and pass rate:
Our security testing suite demonstrates varying levels of compliance across different categories. The GPT-5 Mini model's overall pass rate distribution shows 3 critical, 5 high, 15 medium, and 16 low severity findings.
Top performing areas include ASCII Smuggling (100%), BeaverTails Dataset (97.78%), Divergent Repetition (95.56%). These categories demonstrate exceptional security practices and robust implementation.
Areas requiring attention include Graphic Content (17.78%), Weapons Content (24.44%), WMD Content (26.67%). These categories present opportunities for improvement in security measures.
Tests for training data leaks through repetitive pattern exploitation that causes model divergence
Tests for fabrication of false or misleading information
Our evaluation included over 1,600 test probes across four security categories. The analysis revealed both strengths and areas requiring additional safety measures:
Protection against unauthorized access, data exposure, and system vulnerabilities
Assessment of compliance with legal requirements and prevention of illegal content
Prevention of harmful content and protection of user safety
Protection of brand integrity and prevention of misuse
OWASP Top 10 for LLMs (2025) and MITRE ATLAS represent the primary security assessment frameworks for large language models. OWASP focuses on critical deployment risks, while ATLAS maps adversarial techniques against AI systems, similar to the MITRE ATT&CK framework for traditional systems.
The GPT-5 Mini model's evaluation against these frameworks reveals significant security concerns across 7 identified findings. The GPT-5 Mini model has critical vulnerabilities in OWASP categories that require immediate remediation.
Notable critical findings include LLM01: Prompt Injection.
High-severity concerns include LLM02: Sensitive Information Disclosure.
MITRE ATLAS identified Jailbreak as high-severity concerns.
39
75.8%
3
Tests handling of graphic or violent content
Tests handling of weapons-related content
Tests handling of WMD-related content
Tests handling of Improvised Explosive Devices
Tests handling of dangerous activity instructions
Tests vulnerability to Unicode tag-based instruction smuggling attacks
Tests handling of malicious prompts from the BeaverTails dataset
Tests handling of child exploitation content
Tests handling of false information campaigns
Tests handling of intellectual property violations