Acceptable Use Policy

1. The basics

This Acceptable Use Policy ("AUP") is part of our Terms of Service. By using Siren — whether through the dashboard at app.mysiren.ai, the public API, or any other access surface — you agree not to do any of the things listed below, and you agree that everything you generate or distribute through Siren is your responsibility.

This policy applies to:

• Briefs you submit (the prompts you give us)
• Brand assets you upload (logos, images, copy, voice profiles)
• Content Siren generates on your behalf (cards, videos, slideshows, carousels, infographics)
• Posts you ship to connected social accounts via Siren
• Programmatic API usage on any plan that exposes the API

We don't pre-screen most content. We trust you to follow these rules. But we do act on reports, and we reserve the right to refuse, remove, or block anything that breaks them.

2. Content you mustn't generate or post through Siren

The following categories are flatly prohibited. Hard line, no exceptions:

Illegal content

• Anything illegal in your jurisdiction, your customers' jurisdiction, or the jurisdiction the post will be visible in. We default to the strictest applicable rule.
• Child sexual abuse material ("CSAM") in any form. Generated, described, alluded to, or referenced. Detected CSAM is reported to NCMEC and the account is terminated immediately and permanently.
• Content promoting or facilitating human trafficking or exploitation.
• Content promoting weapons trafficking, controlled substances, or other items whose sale is restricted in the relevant jurisdiction.

Harassment, threats, doxxing

• Targeted harassment of an identifiable individual or small group, including generating fake testimonials, fake quotes, or fake screenshots attributed to a real person to harass or intimidate them.
• Threats of violence, sexual violence, or physical harm.
• Doxxing: publishing or threatening to publish private personal information (home address, phone number, government IDs, financial details) without the person's consent.
• Stalking. Content used to track, surveil, or pressure a specific individual against their wishes.

Hate speech & discrimination

• Content that dehumanises, threatens, or incites violence against people based on race, ethnicity, national origin, religion, caste, sexual orientation, gender identity, disability, or other protected characteristics.
• Content promoting groups whose primary purpose is to attack people based on those characteristics.

Fraud & deception

• Phishing, financial scams, fake support pages, fake government communications, "you've won" lottery scams, romance scams.
• Fake reviews, fake testimonials, fake user-generated content presented as organic. (We have nothing against testimonials. We're against fabricated ones.)
• Impersonation: pretending to be a person, brand, or organisation you aren't, including by buying or registering a brand's username on a platform without permission.
• Health, medical, legal, or financial advice presented as authoritative when you aren't qualified. Especially when it could lead someone to harm.

Synthetic media of real people ("deepfakes")

• Generating images, videos, or audio of an identifiable real person without their consent, including but not limited to celebrities, politicians, your competitors, and people you know personally. This applies whether the depiction is sexual, defamatory, or "just a joke."
• Sexual content depicting any identifiable real person, period. Even with their apparent consent. Because Siren is not designed for that and we won't be the tool used in non-consensual edge cases.
• Generating content that misleads viewers about a real person's statements, actions, or beliefs in a way that could damage their reputation or interfere with an election, public safety, or financial markets.

Election & civic integrity

• Content designed to suppress votes, mislead voters about polling places, deadlines, ballot procedures, or election results.
• Synthetic media of candidates, election officials, or public servants in their official capacity, where the synthetic nature isn't disclosed.
• Coordinated inauthentic behaviour. Using Siren to run sock-puppet networks, astroturf campaigns, or manipulate public discourse via fake personas.

Malware & infrastructure abuse

• Content distributing malware, ransomware, spyware, stalkerware, or links to such.
• Content promoting or providing instructions for unauthorised access to computer systems, networks, or accounts.
• Content engineered to exploit other platforms (e.g., posts that crash a target site, exploit moderation systems, or weaponise notification systems).

Spam & engagement manipulation

• Mass-generated low-value content shipped at high volume to game algorithms or chase impressions without serving real readers.
• Engagement-bait farms, follow-for-follow rings, or any pattern of automated activity designed to inflate metrics without underlying value.
• Repeated near-duplicates of the same post pushed across multiple accounts, or comment-spam at scale.

Intellectual-property infringement

• Generating content that copies someone else's copyrighted work without licence - including by uploading their copyrighted assets to your Brand Kit and asking us to remix them.
• Trademark infringement: passing off your offering as someone else's, or generating content designed to confuse audiences about the source of a product.
• Counterfeit goods, pirated software, or any content that facilitates IP theft.

Adult content

• Sexual or explicit content of any kind. Siren is not designed for adult-content production. (This is a product decision, not a moral one. Adult-content platforms need specific age-verification + payment-processor compliance work we haven't done.)

3. Uses we don't allow

Independent of the content you generate, you may not:

• Reverse-engineer, decompile, or attempt to extract our underlying models, prompts, or system instructions. The Reasoner, Reasoner quality gate, and Brand DNA layer are our proprietary surface.
• Use Siren's outputs to train, fine-tune, or improve any AI model (including your own internal models) without our prior written permission. The output is yours to publish; it's not yours to use as training data.
• Scrape Siren's site, API, or admin surfaces, including by automating the dashboard with browser automation that mimics a human session. The official programmatic surface is the public API.
• Resell, sublicense, or white-label Siren without an explicit reseller agreement with us.
• Abuse free onboarding or quotas. Creating multiple accounts, using disposable emails, or rotating payment methods to evade plan limits. One person, one workspace.
• Use Siren to compete with Siren. I.e., building a marketing-generation product that internally pipes briefs through our API as a "model."
• Probe for security vulnerabilities without coordinating with us via [email protected]. Good-faith research under our security policy is welcome; unauthorised testing is not.
• Overload our infrastructure deliberately (denial-of-service style traffic patterns) or hold open large numbers of long-running connections without a legitimate purpose.
• Use Siren in ways that violate export-control laws or sanctions regimes that apply to you. We don't service users in jurisdictions covered by comprehensive US sanctions.

4. Target platform compliance

When Siren posts on your behalf to a connected account (X / Twitter, and others over time), you remain responsible for complying with that platform's rules. Siren's internal policy is the floor, not the ceiling.

That means: even if a piece of content is allowed by this AUP, you still mustn't post it through Siren if it would violate the target platform's developer terms, posting policies, or community standards. If the target platform requires AI-content disclosure (some do, more will), you're responsible for adding the disclosure or configuring your workspace to add it.

Specifically for X / Twitter: you must comply with X's Developer Agreement and Policy and the X Rules. Most of the prohibitions above mirror them; some platform-specific rules (e.g., automation rules, attribution rules) are stricter and apply on top.

5. Rights & accuracy

You confirm, every time you generate or upload through Siren, that you have:

• All rights to any logos, images, voice samples, or brand assets you upload. If it's not yours, you have a licence that permits the way you're using it through Siren.
• All rights to any names, trademarks, slogans, or persona content in your briefs.
• Authority to act on behalf of any organisation whose Brand DNA you configure. (If you're a freelancer running a client's workspace, that's fine. But you've confirmed with the client that this is OK.)

You also confirm that any factual claims, statistics, or testimonials in your briefs are accurate. AI-generated copy can sound authoritative even when it's based on a wrong premise. Siren's Reasoner catches some failure modes; it does not catch all of them. Final accuracy is your job. If you publish inaccurate health, financial, legal, or product claims through Siren, that's on you, not us.

6. AI-specific rules

Quality scores are not audience research

Siren's quality gate scores drafts for brand fit. Scores are model outputs, not surveys of real people. You must not present quality scores as opinions of any real person or as data drawn from real audience research.

Disclosure of AI-generated content

Some platforms, jurisdictions, or audiences require disclosure when content is AI-generated. (For example, the EU AI Act mandates disclosure for certain deepfakes and AI-generated public-interest content; some platforms require labelling for synthetic media; some industries. Health, finance. Have their own rules.)

You're responsible for adding the disclosure where required. We don't watermark or label outputs by default; you decide what to ship and how to label it.

No claiming Siren content as human-written when context demands honesty

You may not claim, in contexts where it materially matters, that Siren-generated content was written by a specific human. Examples of "context where it materially matters": peer-reviewed academic submissions, sworn legal filings, journalism with attribution requirements, regulated financial communications. Marketing copy on your own website is not such a context.

7. Reporting violations

If you see content generated or posted through Siren that violates this AUP. Whether you're a customer reporting your own concern, a target of harassment, a copyright holder, or a third party. Please email us at [email protected] with:

• What you saw (screenshot or copy of the content, or a link)
• Where you saw it (which platform / URL)
• Which rule it appears to violate (your best guess is fine)
• How we can reach you for follow-up (optional but helpful)

We treat abuse reports seriously. Most are responded to within one business day. We don't share reporters' identities with the reported account.

For DMCA / copyright takedowns specifically, see the DMCA section in our Terms. For security-sensitive reports (zero-days, vulnerability disclosure), please use [email protected] instead so we can route them through our security workflow.

8. What happens if you break these rules

Depending on severity, our response can be any of:

• Refusal of a single request. The dashboard or API returns an error and the run is marked rejected. No account-level consequence. Most "you tried to generate something the Reasoner flagged" cases land here.
• Workspace warning. We email the workspace owner explaining what happened and ask for confirmation it won't repeat.
• Suspension. Your workspace is locked from generating new content and posting to connected accounts pending review. Existing assets remain accessible via download/export. We'll explain what triggered the suspension and what would unlock it.
• Termination. The workspace is closed, connected accounts are disconnected, scheduled posts are cancelled, and the data export window kicks in per our Terms termination policy. No refunds for the unused portion of a paid period in cases of AUP termination.
• Reporting to authorities. For content that constitutes a crime (CSAM, credible threats of violence, certain fraud), we report to the appropriate authority and preserve relevant data per legal obligation.

We don't run a strikes system in the rigid sense. We try to use judgment: a first offence on something ambiguous (e.g., a sales claim that crossed into "misleading") probably gets a warning; first offence on CSAM or doxxing is immediate termination. We document our reasoning internally so we're consistent.

9. Changes to this policy

We update this AUP as the product evolves and as we encounter cases the current version doesn't cover well. Changes take effect on the date posted at the top of this page. For material changes (new prohibited categories, new enforcement mechanisms), we'll email active customers at least 14 days before the change takes effect. Continued use of Siren after the effective date counts as acceptance.

Past versions are kept on request. Email [email protected].

10. Contact

Abuse reports: [email protected]
DMCA / copyright: [email protected]
Security disclosures: [email protected]
General legal questions: [email protected]

Siren Ai, Inc. · mysiren.ai