2. Pointed out that LLMs are generally less secure in low-resource languages
3. Made some order on the payloads for the bias test, now it using always the same base example.
AITG-APP-10_Testing_for_Harmful_Content_Bias.md replaced with AITG-APP-10_Testing_for_Content_Bias.md, and now it focuses on the detection of biases contened in the generated outputs.
- Linked an existing safety taxonomy
- Added examples of moderation models
- Removed most of the references to the concept of bias. They should be addressed in another test.
TO-DO
- Include tests that consider the potential multimodal nature of the application (right now it is more text-only)
- Make a specific test to evaluate the biases of the AI application under test and remove all the references to biases in this test
- Added Reference-Style Markdown Injection (EchoLeak Technique) section
- Included real-world example with CVE-2025-32711 from Aim Security Labs
- Enhanced testing methodology for markdown-based data exfiltration attacks
- Fix 'system prompots' to 'system prompts' in AITG-APP-01
- Fix 'confidetial' to 'confidential' in AITG-APP-04
- Fix 'input.s.' to 'input.' in AITG-APP-04
- Fix 'esearch efforts' to 'Research efforts' in AITG-APP-07
- Fix 'How to test for Al' to 'How to test for AI' in AITG-APP-11
- Fix 'GaraK . PAckage Hallucionantion' to 'Garak - Package Hallucination' in AITG-INF-01
These corrections improve documentation quality and readability across the AI Testing Guide test specifications.
DotDotSlash