Moderation, Enforcement & Appeals

🏛️ Three-Pillar Moderation System

Parsafy's moderation system combines three distinct but complementary approaches. No single approach is sufficient on its own — together they create a moderation ecosystem that is faster, more accurate, and more contextually sensitive than any one method could be.

🤖 AI & Automated Moderation

Parsafy uses a combination of automated tools that operate continuously across the platform to detect potentially violating content before it reaches a wide audience and to identify known harmful content at upload.

Pre-Publication Review

In high-reach areas of Parsafy — including the Discovery Feed and Parsafy TV — content submitted by users undergoes an initial AI review process before receiving algorithmic distribution. This review checks content against our Content Recommendation Guidelines and flags content that may violate our Community Guidelines for human review before it reaches a wider audience.

Hash-Matching Technology

For the most serious categories of content — particularly child sexual abuse material (CSAM) — Parsafy uses hash-matching technologies including PhotoDNA and Google's CSAI Match to identify known illegal content at upload and remove it before it can be viewed by any user.

Machine Learning Classifiers

Parsafy deploys AI classifiers trained to detect categories of potentially violating content — including hate speech, violent content, spam, and misinformation patterns. Content flagged by these classifiers is queued for human review, with urgency determined by the potential severity of the violation.

Keyword & Pattern Filtering

Keyword filters and behavioral pattern detection help identify spam, coordinated inauthentic behavior, phishing, and other rule-based violations at scale across all platform surfaces, including direct messages and comments.

👁️ Human Review

Parsafy's Trust & Safety team consists of trained specialists who review content flagged by automated systems and reported by users. Human review is essential for decisions that require contextual judgment — cases where the same content might be acceptable in one context and a violation in another.

What Human Reviewers Evaluate

• Context and purpose: Is the content educational, journalistic, artistic, or satirical? Or does it promote, incite, or facilitate harm?
• Intent: Does the context of the post suggest harmful intent, or is the content being used in a protected way?
• Account history: Does the account have a pattern of prior violations, or is this an isolated incident?
• Community impact: Is the content likely to cause harm to specific individuals or to the broader community?
• Applicable policy: Which specific Community Guidelines provision, if any, does this content violate?

Specialist Review for Sensitive Categories

Certain content categories — including child exploitation, terrorism, hateful content, and serious threats — are reviewed by specialized teams with domain expertise in those areas. These reviewers receive training specific to the content type and the legal and policy considerations involved.

Review Timelines

Our goal is to review all flagged content in a timely manner, with urgency proportionate to the severity of the potential violation. Content involving imminent threats to life, CSAM, or terrorism is reviewed immediately as our highest priority. Other flagged content is reviewed in order of severity and volume.

🤝 Community Moderation

Parsafy's community is our most important moderation partner. Our users collectively see far more content than our automated systems and human teams combined — and they have context about accounts, relationships, and behavior patterns that our systems cannot easily detect.

User Reports

Any Parsafy user can report content, accounts, or behavior they believe violates our Community Guidelines. Reports are submitted confidentially — the identity of the reporter is never disclosed to the reported user. All reports are reviewed by our Trust & Safety team. For instructions on how to submit a report, see our Reporting Guide.

Community Feedback Loop

User engagement signals — including "Not interested," content dismissals, and explicit feedback — inform how our recommendation systems distribute content. Content that consistently generates negative feedback signals receives reduced algorithmic distribution even where it does not rise to a Community Guidelines violation. This gives users collective influence over content quality without requiring formal moderation action.

Proactive Community Standards

Beyond formal reporting, Parsafy encourages our community to engage constructively, call out misinformation through Community Notes (see below), and model the kind of discourse they want to see on the platform. Community behavior sets the norms of the platform as much as our policies do.

📝 Community Notes

Community Notes is Parsafy's user-powered tool for addressing contested or potentially misleading information without removing content or suppressing the speaker. It is a core part of our commitment to free expression and community-led moderation.

Community Notes is Parsafy's preferred response to contested information — particularly health claims, political statements, and other topics where reasonable people disagree. It empowers the community to surface accurate context without requiring Parsafy to act as an arbiter of truth. For content that falls below the threshold of a Community Guidelines violation, Community Notes is the primary moderation tool.

⚖️ Violation Severity Tiers

Parsafy uses a risk-based approach to enforcement. Not all violations are equal — the severity of the harm and the account's history of violations are the most important factors in determining the appropriate response. We distinguish between four tiers of violation severity:

⚙️ Enforcement Process — Standard Violations

For standard violations — those that fall into the yellow tier above — Parsafy uses a structured three-step enforcement process designed to be consistent, transparent, and educational:

🎯 Strike System

Parsafy uses a strike system to track patterns of repeated policy violations and to apply escalating enforcement responses proportionate to a user's history of violations.

How Strikes Work

• Each confirmed violation of our Community Guidelines (in the standard tier) results in one strike recorded against the account.
• Each strike is communicated to the user at the time it is recorded, along with an explanation of the violation.
• Strikes have a defined expiration window — strikes do not accumulate indefinitely. Users who go extended periods without violations see their strike count reduced.
• Accumulating a specified number of strikes within the applicable window results in escalating enforcement: first a temporary account restriction, then a longer suspension, and ultimately permanent account disabling.

What Strikes Are For

The strike system is designed to:

• Give users clear, documented notice of policy violations.
• Provide a structured escalation path that is predictable and consistent.
• Distinguish between users who make isolated mistakes and users who repeatedly and deliberately violate our policies.
• Allow users who correct their behavior to return to good standing over time.

What Strikes Are Not For

• Strikes are not applied for content that is removed only from recommendation (green tier).
• Strikes are not applied retroactively — only to violations confirmed after these policies take effect.
• Strikes are not the appropriate mechanism for zero-tolerance or serious violations, which bypass the strike system entirely.

📬 Notice to Users

Parsafy is committed to transparency in our enforcement actions. We do not take unexplained enforcement actions. When we take action against content or an account, we tell you:

• What action we took — content removal, strike, account restriction, account suspension, or account termination.
• Why we took it — the specific Community Guidelines provision or Terms of Service section the content or behavior violated.
• What you can do — whether you can appeal the decision, and how.

Notices are sent to the email address associated with your Parsafy account and may also appear as in-app notifications. It is your responsibility to keep your contact information up to date so you receive these notifications.

↩️ Appeals Process

Parsafy believes in due process. If you believe an enforcement action was taken against your content or account in error — or if you believe our decision did not correctly apply our policies — you have the right to appeal. We review all appeals in a timely, unbiased, diligent, and non-arbitrary manner, and we will correct our decision if we determine our initial assessment was incorrect.

What to Include in Your Appeal

• Your Parsafy username and the email address associated with your account.
• A clear description of the content or action you are appealing and why you believe the decision was incorrect.
• Any context, evidence, or explanation that supports your case — for example, evidence that the content was satire, journalism, or counter-speech that was misidentified as a violation.
• For account appeals: any evidence that violations were the result of automated error or misidentification rather than intentional policy breach.

How We Handle Appeals

Limitations on Appeals

• Zero-tolerance violations — account terminations for CSAM, terrorism, and other zero-tolerance violations are not subject to appeal.
• Repeated appeals — submitting multiple appeals for the same decision, or submitting appeals that contain new information already considered in a prior appeal, may not receive a full secondary review.
• Timeframe — appeals should be submitted within 90 days of the relevant enforcement action. Appeals submitted after this window may not be reviewed.

📊 Transparency

Parsafy is committed to transparency in how we moderate content and enforce our policies. We publish periodic Transparency Reports that include:

• Aggregate volume of reports received by category.
• Enforcement actions taken by violation type and severity tier.
• Appeals received and outcomes.
• Government and law enforcement requests received and how we responded.
• Content removed from recommendation vs. content removed entirely.
• Proactive detection rates vs. user-reported detection rates.

Our Transparency Reports are designed to give our community genuine insight into how our moderation system operates at scale — not just high-level statistics, but meaningful data that allows users and researchers to hold us accountable.

We also maintain a transparent algorithm commitment: the factors that determine what content is recommended and what is deprioritized are documented in our Content Recommendation Guidelines. Users are never left wondering why content appeared or did not appear in their feed without explanation.