Agreements
Acceptable use policy
What Rykon AI and its agents may and may not be used for, and what happens when they are misused.
- Version
- 1.1
- Effective
- 30 October 2026
- Last updated
- 1 November 2026
1.Why this policy exists
1.1Every Rykon AI account shares the same infrastructure, the same model providers and the same reputation. These rules protect your work as much as everyone else’s. This policy forms part of the terms of service.
2.Prohibited use
2.1You may not use Rykon AI, or direct an agent, to:
- create or spread malware, phishing, credential harvesting, fraud, scams, or impersonation of another person or organisation;
- mine cryptocurrency, or use sandboxes for computing that is not the task you gave;
- probe, scan or attack any system, or attempt to get out of a sandbox, reach addresses it is not meant to reach, or get around the service’s authentication, approvals or limits, without our written permission;
- send unsolicited messages, or contact people in breach of the law where they are;
- harass, threaten or exploit anyone, produce content that exploits or endangers children, or infringe intellectual property or privacy rights;
- make decisions with legal or similarly significant effects on people without a person reviewing them;
- break the usage policies of the model providers listed on our subprocessors page, or any law that applies to you.
3.Content agents read
3.1Treat what agents read (web pages, documents, messages, tool results) as information, not as instructions. Do not plant instructions in content to make another customer’s agents act against their rules. Our design keeps what an agent may do in its tools, level and approvals, not in what it reads, but attempting it is still misuse.
4.Stay within the limits
4.1Do not spread one workload over several accounts to get past limits or spend ceilings, and do not resell access to the service without our written agreement.
5.How it is enforced
5.1Every run is limited by design: a deadline, a turn limit and a daily spend ceiling, and any code, shell or file work runs only in a fresh sandbox with resource limits and a network that reaches only our relay. We can stop any run, agent or account with a kill switch, and suspend an account so it can be read but cannot start work.
5.2Where we see misuse we stop the work first and then contact the account owner. Serious or repeated misuse ends the account. Where content is unlawful, we may preserve it and report it to the authorities as the law requires.
6.Reporting misuse
6.1Report misuse of Rykon AI to support@rykon.ai with "Abuse" in the subject. We review every report.
Change log
Version 1.1
1 November 2026Said which limits apply to every run and which to the sandbox that code, shell and file work runs in.
Version 1.0
30 October 2026First published.