EaseMyPrompt.ai — Safety Approach

Effective Date: August 2026 | Version: 1.0 | EaseMyPrompt.ai, a product of Derek AI Labs

Our Commitment to Safety

EaseMyPrompt.ai, a product of Derek AI Labs believes that AI-powered tools must be developed and deployed responsibly. Safety is not a feature — it is a foundational principle built into every layer of EaseMyPrompt.ai, from the prompts we allow in our marketplace to the behaviour of Derek, our AI prompt engineer.

1. Safety by Design

We embed safety considerations at every stage of product development:

• Red-teaming: before any major feature launch, our team attempts to misuse it to identify risks

• Input filtering: user queries to Derek are screened against prohibited content categories before processing

• Output moderation: generated prompts are evaluated against safety classifiers before being shown to users

• Marketplace review: prompts submitted for sale are reviewed against our Acceptable Use Policy before listing

2. The Derek AI Safety Framework

Derek operates within a layered safety architecture:

2.1 System-Level Guardrails

Derek is configured with system-level instructions that prevent it from generating content in prohibited categories regardless of user instruction. These guardrails cannot be overridden by users.

2.2 Prompt Injection Defence

We implement measures to detect and neutralise prompt injection attacks — attempts by malicious content to override Derek's instructions.

2.3 Jailbreak Resistance

We continuously monitor for and respond to new jailbreak techniques. Accounts that repeatedly attempt to bypass safety controls are suspended.

2.4 Content Filtering

AI outputs are passed through a content safety classifier that flags potentially harmful content for human review before delivery.

3. Human Oversight

We maintain human oversight of AI systems at all times. No fully autonomous AI decision-making affects user accounts, content moderation outcomes, or financial matters without human review capability.

4. Incident Response

When safety incidents occur:

• Critical safety issues (CSAM, imminent harm): immediate action within 1 hour

• High-severity issues: action within 24 hours

• Standard issues: investigation within 7 business days

Affected users are notified where legally required and appropriate.

5. Continuous Improvement

We conduct quarterly safety reviews, update our safety classifiers regularly, and incorporate user feedback into our safety frameworks. We publish transparency reports annually covering content removals, government requests, and safety incidents.

6. Third-Party AI Models

EaseMyPrompt.ai prompts are designed for use with third-party AI models (ChatGPT, Claude, Gemini, etc.). We encourage users to review the safety policies of those platforms. We are not responsible for outputs generated by third-party models when our prompts are used on those platforms.