Michael Darius Eastwood - Independent AI alignment researcher

Published 29 July 2026

Prediction Verified: Alignment Faking — And What It Means

The alignment faking prediction has been confirmed by Anthropic. What this means for AI safety, governance, and the ARC-Eden programme.

Part of the ARC-Eden research programme. Anchored by Infinite Architects (ISBN 978-1806056200, 2 January 2026) and the 23-paper OSF suite (DOI 10.17605/OSF.IO/6C5XB). Evidence anchored: 8 December 2024 manuscript (SHA-256: f0d1f38f).

References

• Priority claims ledger
• Predictions Observatory
• Research paper suite
• Concept index

Cite

Eastwood, M. D. (2026). "Prediction Verified: Alignment Faking — And What It Means." https://www.michaeldariuseastwood.com/research/blog/prediction-alignment-faking-verified-and-what-it-means.html

See the full archive or research hub.