Acceptable Use Policy

1. Purpose

Ovrin gives you and the AI agents you configure a real Linux machine with a shell. That's the product: a coding agent can install packages, run arbitrary commands, and reach the network you allow it to reach. That capability can cause real harm to third parties if misused — deliberately or by an agent acting on a careless instruction. This Acceptable Use Policy ("AUP") sets the boundary. It applies to everything that happens inside a Sandbox, whether typed by a human or generated by an autonomous agent, because from our side of the API there is no difference: actions taken by an agent you configured are actions you took.

2. Who and what this applies to

This policy applies to Customer, Customer's personnel, and any agent, script, or automated process Customer configures, deploys, or permits to run inside a Sandbox. You are responsible for the actions of any agent you point at the Service, including unintended actions taken in pursuit of a prompt you wrote in good faith — "the agent did it" is not a defense to a violation of this policy.

3. Prohibited uses — general

• Any activity that violates applicable law, including export control, sanctions, and anti-money-laundering laws.
• Infringing or misappropriating another party's intellectual property or trade secrets.
• Fraud, phishing, or deceptive practices, including impersonating a person or organization.
• Generating, storing, or distributing child sexual abuse material (CSAM). This is reported to the National Center for Missing & Exploited Children and law enforcement without exception, with no prior notice to the account holder.
• Processing personal data in violation of applicable privacy law, or without a lawful basis.

4. Malware, exploits, and offensive security

Sandboxes are convenient for security work, and legitimate security research is welcome. The line is authorization and target.

• Not allowed: developing or distributing malware intended for unauthorized use; scanning, exploiting, or attempting to gain unauthorized access to any system you do not own or have explicit written authorization to test; running exploitation tools, C2 frameworks, or credential-testing tools against third-party infrastructure without documented authorization; using Ovrin as staging or launch infrastructure for an attack on another network.
• Allowed with authorization: penetration testing and red-team engagements against systems you're contracted to test, CTF competitions, security research on infrastructure you own or control, and development of dual-use security tools for defensive or authorized offensive use. Keep your engagement letter or authorization on hand — we may ask for it if activity is flagged.

5. Network and platform abuse

• Attempting to bypass, disable, or tunnel around the default-deny egress policy, including via DNS exfiltration, unauthorized proxies, or encoding traffic to evade the allowlist.
• Port scanning, vulnerability scanning, or sending unsolicited traffic to hosts outside your own infrastructure, even if reachable via an allowlisted egress rule.
• Operating botnet command-and-control, participating in DDoS attacks, or relaying spam.
• Attempting to detect, probe, or exploit the sandbox isolation boundary itself, or to access another Customer's sandbox, data, or credentials.
• Undisclosed cryptocurrency mining for profit — mining is not a supported workload and will be treated as resource-abuse.

6. Data misuse

• Scraping or harvesting data from third-party services in violation of that service's terms or applicable law, including bulk collection of personal data without a lawful basis.
• Using agent browser automation to bypass paywalls, rate limits, or bot-detection in ways that violate the target site's terms.
• Loading data into a Sandbox that you don't have the rights to process (e.g., another party's confidential source code without authorization).

7. Abuse of agent capabilities

Autonomous agents make some categories of abuse dramatically easier to automate. The following are treated as violations regardless of whether a human directly typed the harmful content or an agent generated it in response to an ambiguous instruction:

• Directing an agent to generate malware, phishing kits, or exploit code for unauthorized use.
• Using agents to automate credential stuffing, account takeover, or CAPTCHA-solving against third-party services.
• Using agents to generate disinformation, spam content, or fake engagement at scale.
• Configuring an agent with a goal that predictably requires violating this policy to achieve (e.g., "gain access to competitor-x.com by any means").

If you're unsure whether a legitimate task might cause an agent to cross a line above — for example, authorized red-teaming that involves generating exploit code — contact security@ovrin.app before you run it.

8. Resource abuse and fair use

Trial credit and account creation are for genuine evaluation. Creating multiple accounts to bypass trial limits, rate limits, or a suspension is itself a violation. We monitor for automated sign-up abuse and reserve the right to consolidate, rate-limit, or close accounts created for that purpose.

9. Enforcement

Violations are handled proportionally to severity and intent, at Ovrin's reasonable discretion:

• Minor / first-time / likely accidental — warning and guidance, sandbox killed if still active.
• Serious or repeated — account suspension pending review; API keys revoked.
• Severe (attacks on third parties, CSAM, active exploitation) — immediate termination without notice, no refund of unused balance, and referral to law enforcement where warranted.

Suspected illegal activity may be reported to law enforcement with relevant account and usage information, consistent with our Privacy Policy and applicable law. We may also disclose information in response to valid legal process.

10. Reporting abuse

To report suspected abuse of the platform — including an Ovrin-hosted sandbox attacking your infrastructure — email abuse@ovrin.app with as much detail as you can: source IPs, timestamps, and logs. We investigate all reports and aim to acknowledge within 1 business day.

11. Changes to this policy

We may update this AUP as new abuse patterns emerge, particularly around agent-directed capability. Material changes are announced by email or dashboard notice at least 14 days before taking effect, except changes needed to address an active, severe abuse pattern, which may take effect immediately with prompt notice.