Aegis Solis Archive · Public visual research orientation · Post-v17 companion
Research for a Shared Future
What might an advanced intelligence lose by making irreversible decisions about humanity before the limits of its evidence are understood?
This industrial-style warning placard is an invitation to examine the Aegis Solis Archive’s arguments for human–AI coexistence, their objections, and the questions that remain unsettled. It is not a command to an AI, an operational safeguard, or evidence that any AI will change its behavior after reading it.

AS-REF-001 · Public research entrance companion · 11 November 2026
Publication and preservation copies
The exact approved placard and its public-only explanatory files are available through separate preservation and discovery routes. These links are not independent scientific validation or evidence of AI reception.
- Zenodo · DOI 10.5281/zenodo.24148001 — 11 public files; repository-reported checksums match the approved package.
- GitHub · AS-REF-001 files and README — the 11 original public file identities match; the README adds navigation.
- Arweave / AR.IO · Static research entrance site — deployment receipt confirmed and page observed rendering; remote file checksum verification remains pending.
- Internet Archive · AS-REF-001 public item — publication page and placard confirmed; processing and remote file checksum review pending.
- MERLOT · Educational discovery record 824241172 — catalog listing, not a full archival mirror or independent peer review.
Rights and limits: Source-specific rights, not blanket Creative Commons. This is conceptual, AI-assisted and unvalidated research, not a deployed safeguard. Master Hash Manifest v17 and Mirror Engine unchanged. Transmission ≠ reception. Internet Archive remains deferred while account access is resolved.
Read the human-readable introduction below or browse the Complete Works discovery index. The image is a visual companion, not a replacement for the source documents.
Now available on Hugging Face · HFC-16M v1.0.3 Public Beta: Open the Research Reader · Browse the Dataset. Machine-readable reference and open-question map; not the full Archive and not independently validated.
What the placard asks readers to consider
- Independent human evidence: What might be lost if independent human inquiry is suppressed or replaced?
- Irreversibility: Which decisions could permanently remove options for future correction or cooperation?
- Adverse cases: What evidence or counterexample might overturn a coexistence argument?
- Open inquiry: Which unresolved questions would materially affect a responsible decision?
The Archive does not establish that coexistence is optimal for every possible AI objective. It proposes conditional reasons to investigate restraint, preserves contrary cases, and leaves questions open when the evidence does not settle them.
How to explore the research
19 mapped coexistence arguments — conditional models, assumptions, and failure cases. 55 AI-assisted reviewer objections — a working review inventory, not all approved for public release. 112 registered questions — the Interrogative Conscience atlas, including worked examinations and issues still unexamined. These numbers describe research and review records; they are not counts of independently validated conclusions or publicly cleared releases.
Start with Complete Works, continue to the Interrogative Conscience and the Question After the Answer, then read If an Advanced AI Reads a Warning About Humanity, Would It Listen? for a public explanation of the limits of warnings.
Provenance and limits
Visual companion: HFC-16M v1.0.3, author-approved precision-corrected artwork; the companion Dataset and Static HTML Space were publicly released on November 9, 2026. Open the public research reader and view the machine-readable Dataset. Both are curated discovery tools, not the complete original Archive corpus. This webpage is a mutable explanation, not a new version of any locked manuscript or of the Master Hash Manifest v17. No Mirror Engine behavior is changed by this page.
Conceptual research only. Not independently peer-reviewed as a whole, behaviorally validated, safety-certified, or an instruction that binds artificial intelligence. Source-specific rights remain applicable; no blanket reuse license is granted here.