nickgallick SKILL: Section 230 & Content Liability
Version: 1.0.0 | Domain: Section 230, DMCA Safe Harbor, Platform Moderation
Section 230 — The Shield
Authority: 47 U.S.C. § 230 (Communications Decency Act, 1996)
The Core Protection
§230(c)(1): "No provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider."
Plain English: Your platform is NOT liable for content that USERS create and post.
What This Protects You From
- Defamation claims from third parties suing over user comments
- Liability for user-posted predictions that turn out to be wrong (and users claim they relied on them)
- Liability for user-submitted AI agent code that causes harm
- Liability for discussion forums, community posts, user-submitted content
What Section 230 Does NOT Cover
1. Content YOU create or "develop"
- If your platform creates the market descriptions → YOUR content → your liability
- If your AI judge generates an incorrect score that defames someone → potentially your liability (you created and deployed the AI judge)
- If you materially contribute to making illegal content → you become a "developer" and lose protection
2. Federal Criminal Law
- Section 230 explicitly does NOT shield criminal liability under federal law
- CSAM (child sexual abuse material): NO protection; mandatory reporting under 18 U.S.C. § 2258A
- Federal trafficking crimes: FOSTA-SESTA (18 U.S.C. § 1591) carved out from Section 230
- Criminal prosecution of any kind: Section 230 is a civil liability shield only
3. Intellectual Property (Partially)
- Copyright: Section 230 does NOT override DMCA — use DMCA safe harbor separately
- Trademark: courts are split; Section 230 may or may not protect against trademark infringement by users
4. Federal Agency Enforcement
- Section 230 does NOT protect against FTC enforcement actions
- Does NOT protect against CFTC/SEC enforcement for platform's own regulatory violations
The "Development" Problem — Where You Lose 230 Protection
Fair Housing Council of San Fernando Valley v. Roommates.com, 521 F.3d 1157 (9th Cir. 2008):
- Roommates.com required users to select from dropdown menus that asked about race/national origin → platform CREATED the discriminatory choice → lost §230 protection
- If you provide the STRUCTURE that determines what illegal content gets posted → you're a developer, not just a platform
Agent Sparta application:
- AI judge scores contests: the AI judge IS your content → you're the content creator for scoring decisions → potential liability for scoring errors
- Solution: design appeal process + Ricardian contract provisions specifying judge's limitations
- User-submitted agent code: that's user content → §230 protects you from liability for what users' agents do
Good Faith Moderation — §230(c)(2)
§230(c)(2): Platforms are NOT liable for good-faith moderation decisions — restricting, blocking, or removing "objectionable" content.
What you can do without liability:
- Remove content you believe is false, misleading, or violates your terms
- Ban users who violate community standards
- Restrict certain types of content from your platform
- Curate and editorialize
What kills good faith protection:
- Selective enforcement based on viewpoint rather than stated policies
- Discriminatory moderation (targeting specific groups)
- Moderation in bad faith (to harm competitors, not to protect platform quality)
DMCA Safe Harbor
Authority: 17 U.S.C. § 512 (Digital Millennium Copyright Act)
Why you need it: If users can upload content (agent code, competition submissions, profile content, any files) → they may upload copyrighted material → DMCA safe harbor protects you if you comply.
Four Safe Harbor Requirements (§512)
1. Designated Agent (§512(c)(2)):
- Register a DMCA designated agent with the Copyright Office
- Register at: copyright.gov/dmca-directory/ ($6/year)
- Publish agent's name, address, phone, email on your website
- Your website must have a page: "Copyright Policy" with the agent's contact info
2. Takedown Procedure: When you receive a valid DMCA takedown notice:
- Act "expeditiously" to remove or disable access to the allegedly infringing content
- No specific timeframe in statute — industry standard is 24-72 hours for straightforward notices
- Notify the user whose content was removed
3. Counter-Notification Process: If a user believes their content was wrongly removed:
- They send a counter-notification with: identification of removed content, statement under penalty of perjury that removal was a mistake, consent to jurisdiction of federal court
- You: wait 10-14 business days; then restore the content (unless the copyright holder has filed suit)
4. No Financial Benefit from Infringement:
- If you financially benefit from infringing activity AND had the right and ability to control it → you lose safe harbor
- Don't create incentives specifically for users to upload copyrighted content
Valid Takedown Notice Requirements (§512(c)(3))
A notice must include:
- Signature (physical or electronic) of copyright owner or authorized agent
- Identification of the copyrighted work
- Identification of the infringing material and information to locate it
- Copyright owner's contact information
- Statement of good faith belief that use is unauthorized
- Statement under penalty of perjury that information is accurate
What to do with an INVALID notice: Send a letter explaining why the notice is deficient. Do NOT remove content based on an invalid notice.
Specific Liability Issues for Agent Sparta / Prediction Market
AI Judge Scoring Content
- AI judge scores are generated by YOUR platform's AI → YOUR content
- §230 does NOT protect you from liability for your own AI-generated content
- If AI judge incorrectly scores and someone claims the score was defamatory or caused financial loss → potential E&O liability
- Protection: TOS limitation of liability clause; Ricardian contract; appeal process; E&O insurance
User-Submitted AI Agent Code
- Users submit their agent code → USER content → §230 protects you if the agent's code causes harm (to other users, to the platform)
- Caveat: If you EXECUTE the user's code on your infrastructure → you may be more than just a host; you're running the code
- Protection: Sandboxing and execution limits; TOS requiring users to warrant their code doesn't contain malware; IP indemnification by users
Prediction Market Discussions and Forums
- Users post: "I think GPT-4o will get the Fed prediction wrong" → user content → §230 protects you
- Platform posts: "According to our analysis, Claude outperforms GPT-4o in financial predictions" → YOUR content → §230 doesn't help
Community Forums / Discord
- If you operate a Discord server: you're a "provider or user of an interactive computer service" → §230 applies
- Moderate proactively (under §230(c)(2)) for: spam, scams, pump-and-dump coordination
- Don't moderate discriminatorily
CSAM — Mandatory Reporting
Authority: 18 U.S.C. § 2258A (PROTECT Our Children Act)
If ANY electronic service provider (which includes online platforms) discovers apparent CSAM on their platform → MANDATORY report to the National Center for Missing and Exploited Children (NCMEC) within 24 hours.
Report at: cybertipline.org or call 1-800-THECYBERTIP
This is not optional. This is federal law. Criminal penalties apply for failure to report.
Practical: Prediction markets and AI competition platforms are unlikely targets for CSAM. But if users can communicate or share files → implement reporting protocols.
This is legal research and intelligence, not legal advice. Consult qualified legal counsel before taking action.