← Back to Home
🛡️ Governance & Safety Framework

AI Safety & Risk Policy

Effective Date: August 21, 2026

At Work Less AI (a product of Z Media Technologies Private Limited), safety, security, and ethical alignment are foundational to everything we build. As we empower businesses with autonomous AI workers, we maintain a proactive framework to manage AI risk, prevent misuse, and ensure human oversight across all automated operations.

1. Our Commitment to Responsible AI

We are dedicated to building autonomous AI agents that operate transparently, reliably, and within safe boundaries. Our platform integrates state-of-the-art model alignment, continuous safety monitoring, and strict access controls to prevent harmful or unintended behaviors.

2. Core AI Safety Principles

• Human Alignment & Intent: Autonomous AI agents strictly follow user-specified objectives and system instructions without drifting into unauthorized actions.
• Human-in-the-Loop (HITL) Controls: Users maintain full control over agent workflows. Critical actions (such as financial transactions, deployment commands, or sensitive email dispatches) can be configured to require explicit human approval before execution.
• Hallucination Management & Verification: Agents utilize retrieval-augmented generation (RAG) and self-checking loops to verify facts against ground-truth data sources, minimizing hallucinations.
• Data Privacy & Isolation: Customer prompts, proprietary data, and execution logs are isolated per tenant. Customer data is never used to train foundational public models without explicit opt-in consent.

3. Risk Management & Technical Guardrails

To defend against emerging AI security risks, Work Less AI implements multi-layered safeguards:

• Sandboxed Execution Environments: Development and coding agents run code inside isolated, ephemeral containerized environments with restricted network privileges.
• Prompt Injection & Jailbreak Defense: Real-time input filters analyze user prompts and incoming third-party data to detect and block prompt injection, jailbreak attempts, and instruction override attacks.
• Rate Limiting & Cost Caps: Automated safeguards prevent runaway execution loops, capping API consumption and task depth to protect against unintended resource exhaustion.

4. Prohibited Uses & Misuse Prevention

To uphold systemic safety, Work Less AI strictly prohibits the use of our autonomous agents for any of the following high-risk activities:

• Generating or distributing malware, spyware, exploits, or automated cyberattack scripts.
• Automated phishing, spam campaigns, or deceptive social engineering operations.
• Creating non-consensual synthetic media, deepfakes, or hate speech.
• Bypassing authentication controls or accessing unauthorized third-party infrastructure.
• Engage in illegal financial manipulation or unauthorized medical/legal advice generation.

5. Auditability & Incident Response

All agent decisions, tools invoked, and execution logs are recorded in audit trails accessible to account administrators. In the event of a suspected AI safety incident or anomalous agent behavior, our automated emergency brake allows administrators to instantly terminate running agents.

6. Vulnerability & Safety Reporting

We welcome reports from AI safety researchers, security analysts, and users. If you discover an AI alignment issue, security vulnerability, or prompt injection vulnerability on our platform, please report it to us immediately at support@iworkless.app.