Acceptance
By using Jarvis (the "bot" or "service"), you agree to these Terms. You must be at least 13 and meet the minimum age required to use Discord in your country. If you cannot legally agree to these Terms on your own, a parent or legal guardian must agree on your behalf. If you use Jarvis for an organization or install him in a server, you confirm that you are authorized to do so. If you don't agree, please don't use Jarvis.
These Terms are between you and the independent Jarvis operator, not Discord. They do not replace or modify Discord's terms or policies, which control if there is a conflict concerning Discord or its APIs. Nothing here limits mandatory consumer or data-protection rights that applicable law does not allow us to waive.
What Jarvis does
Jarvis is a Discord assistant. He provides AI chat (with keyword-based memory recall and reply-aware conversations), voice conversations, music playback, AutoMod, server stats, and server-engagement tools such as a leveling system. Some responses are generated by external AI services (see Third-party services for the current list).
Acceptable use
Follow Discord's Terms of Service, Privacy Policy, Community Guidelines, Developer Terms, and Developer Policy. Don't use Jarvis for spam, harassment, scraping, evading permissions, credential collection, adult content in non-age-restricted spaces, illegal activity, or anything that would make a Discord app unsafe or non-compliant.
Data
Jarvis processes your messages when you talk to him - mentions, commands, context-menu actions, or voice chat. Regular one-to-one AI chat DMs are disabled to prevent abuse, but some supported commands and context-menu actions may be available in Discord private, group, or user-install contexts. If you reply to someone's message while talking to Jarvis, he reads that single replied-to message for context (and skips it if that author used /opt out). When message-content features are enabled, Jarvis may also read a small recent window from the same channel to follow the live conversation; opted-out users are skipped. A few opt-in server features a server admin can switch on - such as the counting game or voice-message transcription - read messages or attachments only in the channels where those features apply. In voice channels, listening only starts after someone explicitly runs /voice; /play alone does not enable it. Once active, he only processes opted-in users. In solo voice calls, he listens continuously without requiring a wake word for lower latency. In multi-user voice calls, he usually reacts to his name (wake word), but may keep processing short follow-up speech from the active speaker during an ongoing exchange. Raw audio is not intentionally recorded or stored; it is processed live for speech-to-text and then discarded.
When Jarvis generates a response, your message is sent to an AI provider (whichever is available - see Third-party services). The resulting conversations are saved, encrypted at rest, against your account so context can persist between sessions. These memories are scoped to you, not to a server. Server-specific settings (channel locks, role gates, AutoMod rules, economy, and other guild config) are kept separately and scoped to that one server.
You're in control. Run /opt out to wipe stored conversation memories tied to your account - because they are account-scoped, this clears them across every server at once and stops future memory retention. /clear resets your current conversation context. Operational logs, abuse records, portal sessions, and server-scoped feature records are handled under the retention rules in the Privacy Policy. We never sell your data or use it for marketing. We store only what is required for functionality and safety, with strict retention limits and access controls.
Abuse-detection notice: Your stored memories may be read by our automated abuse-prevention systems and processed by external tools that facilitate abuse detection. If you do not agree, run /opt out and discontinue use of Jarvis. Removing the bot from one server does not delete your memories or stop this processing while you keep using Jarvis elsewhere - only /opt out together with discontinuing use does.
To keep the service functional for everyone and prevent abuse, we keep simple per-server and per-user request and AI token counters for up to 30 days, after which they are auto-deleted. These counters exist solely to ensure service functionality for all users.
The unabridged version is in our Privacy Policy.
Uptime & changes
Jarvis may go down for maintenance, hit rate limits, or gain new features without advance notice. The service evolves over time.
AI accuracy, persona & outputs
AI responses can be wrong, incomplete, strange, hostile, sarcastic, offensive, or absurd. Jarvis is intentionally configured with an aggressive fictional AI persona, so responses may sound unhinged or darkly humorous. This does not mean we endorse, verify, recommend, or intend any statement, joke, link, GIF, or action the model produces. Use your own judgment. Jarvis is provided "as is," with no warranties or guarantees.
Limitation of liability
To the fullest extent the law allows, we are not liable for any indirect or consequential damages arising from your use of Jarvis.
Your responsibility for content & conduct
You are solely responsible for the content you submit through Jarvis (messages, prompts, commands, attachments, voice input, context-menu actions like Quote and Clip) and for the conduct that flows from how you use the bot in your server. Automated controls may screen user inputs, stored memories, tool requests, and generated output for certain high-risk content, abuse patterns, permission bypass attempts, or operational failures. Depending on the result, Jarvis may refuse, replace, shorten, or discard a response or tool request. These controls do not inspect or catch every harmful output, and no provider or automated system is perfect. You remain responsible for sharing, amplifying, screenshotting, or otherwise propagating output the bot generates.
You agree not to use Jarvis to: (a) attempt to elicit content involving minors in any sexual, romantic, or exploitative context; (b) attempt to elicit slurs, self-harm directives, threats, or targeted harassment toward real identifiable people; (c) coordinate doxxing, stalking, or harassment campaigns; (d) generate content intended to defame, impersonate, or sexually depict real named individuals; (e) bypass, jailbreak, or otherwise circumvent the safety measures described above; (f) reverse-engineer, scrape at scale, or abuse rate-limits; (g) use the bot in a way that violates Discord's Terms of Service or Community Guidelines, or any law applicable to you; (h) trick the AI into using tools to access private channels, audit logs, memory, documents, moderation records, or other data you are not authorized to access; (i) use GIF, media, search, or tool features to locate or distribute prohibited, sexual, exploitative, hateful, violent, harassing, or infringing content.
We may temporarily restrict, permanently revoke, and report to Discord any user or server account engaged in the conduct above, at our discretion and without prior notice where reasonably necessary for safety, legal compliance, or service protection. Server owners, administrators, and moderators are responsible for configuring permissions and moderating their servers, including use of the bot's Make it a Quote, Clip Message, and similar features. Moderators with Manage Messages permission can remove bot output using Discord's standard message controls.
If a third party (including Discord) brings a claim against us arising out of your use of Jarvis in violation of these Terms, you agree to indemnify and hold us harmless to the fullest extent allowed by law.
Autonomous & agentic actions
Jarvis has "tool calling" capabilities. During a user-initiated interaction, an AI model may select from tools exposed for that context, including certain server-management or moderation tools. A selected tool must still pass Jarvis access checks and Discord's own user and bot permission checks; sensitive categories are restricted, rate-limited, and recorded in a redacted audit trail. Granting Jarvis a Discord permission does not authorize every user or every conversation to exercise it. Models can still misunderstand a request or choose the wrong permitted tool, so server owners should grant only the permissions they need, configure an audit channel, and review sensitive actions. Authorized administrators and moderators remain responsible for requests they intentionally direct Jarvis to perform, while we remain responsible for operating the service's safeguards.
Leaving
You can remove Jarvis from your server at any time.
Third-party stuff
Jarvis routes AI responses through multiple external AI providers (and voice transcription/synthesis through NVIDIA NIM); the current list, transfer information, and provider privacy policies are disclosed in our Privacy Policy. Each provider operates under its own terms. We select configurations that minimize retention and provider training where those controls are available, but we cannot override a provider's published terms.
GIF and media features may use third-party services such as Klipy, Tenor, and GIPHY, or URLs previously posted by Discord users. Their libraries contain user-uploaded content, may change or disappear, and may include material we did not create, review, or control. Third-party providers are responsible for their own services and content policies, but no provider or automated filter can guarantee that every GIF is accurate, available, non-infringing, or appropriate for every server. Server staff must remove unwanted bot output using Discord moderation controls.
Fair use & abuse prevention
We monitor usage patterns and may temporarily restrict or permanently revoke access for excessive, automated, tool-abusive, non-compliant, or otherwise harmful use, at our discretion and without prior notice. Prohibited abuse includes probing hidden tools, repeatedly attempting forbidden tool calls, trying to dump memories or private channel data, evading rate limits, using the bot as a scraping proxy, or using it to violate Discord's Terms, Community Guidelines, Developer Terms, or Developer Policy. Server admins and moderators can also pause Jarvis on their server or for specific members at any time.
Retention after removal
When Jarvis is removed from a server, we retain limited server metadata and safety records for 30 days for abuse investigation and service protection. Depending on enabled security features, those records may include server identity, aggregated usage counters, configuration, channel/role/permission metadata, AutoMod metadata, invite or webhook metadata, scheduled-event metadata, selected rules/system/announcement messages, and member/role metadata. After 30 days, removal-scoped records are deleted. That server's own config stops applying. Your personal memories are account-scoped (see Data above): they survive the removal and are erased everywhere at once with /opt out.
Contact
Discord: https://discord.com/invite/ksXzuBtmK5 · Email: dev@jorvis.org
Updates
These Terms may change as the service or law changes. We will update the effective date and, where required by law, provide reasonable notice of material changes. Continuing to use Jarvis after the updated Terms take effect constitutes acceptance, except where applicable law requires a different form of consent.