Acceptable Use Policy

The mission of Model Mesh is to provide developers, researchers, and enterprises with accessible, resilient, and high-performance foundation model routing. To safeguard our community, our infrastructure, and the general public, all users of the Model Mesh API must adhere strictly to this Acceptable Use Policy (“AUP”).

By creating an account, invoking any model endpoint, or contributing to the capacity pool, you agree to comply with these standards in all jurisdictions where you operate.

You may not use Model Mesh services, directly or indirectly, to develop, host, or execute applications that involve:

• Malicious Cyber Operations: Creating, optimizing, or distributing malware, ransomware, exploits, automated scanning tools for unconsented penetration testing, credential stuffing scripts, or command-and-control backdoors.
• Illegal Content & Exploitation: Generating, processing, or disseminating child sexual abuse material (CSAM), suicide or self-harm encouragement, human trafficking coordination, or illegal drug/contraband trafficking.
• Weapons & CBRN Threats: Providing actionable instructions or simulations for the design, synthesis, or deployment of chemical, biological, radiological, or nuclear (CBRN) weapons or dangerous explosives.
• Non-Consensual Deepfakes & Impersonation: Generating synthetic media, cloned audio voices, or synthetic imagery designed to impersonate real persons without authorization, create non-consensual sexual imagery, or defraud individuals.
• Fraud, Deception & Social Engineering: Automating phishing emails, financial scams, identity theft, or deceptive impersonation of banking or government institutions.

Model Mesh enforces strict guardrails to prevent resource monopolization and infrastructural degradation:

Because Model Mesh orchestrates traffic to foundational model providers, you must also respect the published acceptable use terms of the underlying model you invoke (including OpenAI Usage Policies, Anthropic Usage Policy, Google Generative AI Prohibited Use Policy, and open-source model licenses).

Model Mesh implements dynamic upstream error translation. If an upstream provider rejects your request due to safety or policy violations, the API returns a structured HTTP 400 error detailing the policy flag.

Model Mesh welcomes benign security research and red-teaming aimed at enhancing AI model safety and finding systemic vulnerabilities. To ensure protection under our Safe Harbor guidelines:

• Do not access or attempt to access other users' accounts, keys, or private data.
• Do not cause degradation or denial of service to Model Mesh production infrastructure.
• Provide our security team with a reasonable window to remediate any discovered vulnerability before public disclosure.

Model Mesh uses automated heuristic rate monitors and upstream policy telemetry to detect anomalous patterns. In the event of a suspected or confirmed violation:

• Immediate Key Suspension: Abusive API keys may be automatically terminated without prior warning.
• Account Closure: Repeated or egregious violations result in permanent account termination and forfeiture of prepaid balances.
• Legal Escalation: In cases of severe cybercrime, child endangerment, or national security threats, Model Mesh cooperates fully with law enforcement authorities.

To report suspected misuse, abuse, or safety violations originating from or targeting the Model Mesh platform: