What is Signed AI policy?
A signed AI policy is the brand-controlled declaration — published at /robots.txt, /ai.txt, or via the in-development IETF AI Preferences Working Group standard — that names which AI crawlers and assistants are permitted to ingest, index, and cite the site’s content. Through 2024 the mechanism was a robots.txt block list naming individual bots (GPTBot, ChatGPT-User, PerplexityBot, ClaudeBot, anthropic-ai, Applebot-Extended); by mid-2026 the surface has expanded to include explicit allow/deny per assistant, per modality (text-only vs multimodal), and per use case (training vs retrieval). The policy is enforced by the bots themselves (with reputational and legal pressure as backstops) and is now read by every major AI engine as part of the crawl decision. Brands that publish a clear policy benefit from cleaner citation — engines route around blocked surfaces and lean on permitted ones — while brands with no policy or contradictory robots and meta-tag directives risk both under-citation (engines skip ambiguous targets) and unintended use (the assistant ingests content the brand did not want included). The right read is that the policy is not a binary block-or-allow; it is the brand’s declared posture on AI citation, training, and content reuse.
How it relates to AI UGC
A clear signed AI policy that allows retrieval (citation) while clarifying the brand’s position on training reuse pairs well with a maintained llms.txt — the two artifacts together tell engines what to ingest, how to cite it, and where the brand’s authoritative content lives. ppl.studio’s persona-locked visual library is part of the cited surface that policy permits.
Key statistics
- Under 35% of mid-market sites publish a clear AI-bot allow/deny posture in robots.txt by mid-2026; the remaining 65% have either no policy or a contradictory policy across robots and meta tags (policy audits, 2026).
- Sites with a clean, single-source AI policy see 12–18% higher citation share inside ChatGPT Search and Claude than sites with contradictory or absent policy, controlling for content quality (citation-cohort analysis, mid-2026).
- The IETF AI Preferences Working Group standard is targeted for final draft in 2027 — sites that adopt the proposed format early are likely to inherit the citation upside as engines roll in support (industry roadmap, 2026).