Skip to content
Longterm Wiki

Encoded Reasoning Detection

AI Controlemerging

Detecting when AI systems use steganography, hidden channels, or encoded communication to circumvent oversight.

Organizations
2
Key Papers
1
Grants
5
Total Funding
$497K
Cluster: AI Control
Parent Area: AI Control

Tags

steganographyhidden-communicationoversight-evasion

Grants5

NameRecipientAmountFunderDate
Meridian — Avoiding Encoded Reasoning in LLMsMeridian$245KCoefficient Giving2025-06
University of Maryland — Study on Encoded Reasoning in LLMsUniversity of Maryland$218KCoefficient Giving2025-06
6-month salary for research into preventing steganography in interpretable representations using multiple agentsHoagy Cunningham$20KLong-Term Future Fund (LTFF)2022-10
3-month stipend and cloud credits to research AI collusion mitigation strategies and develop secure steganographyMikhail Baranchuk$13KLong-Term Future Fund (LTFF)2024-04
Compute for experiment about how steganography in large language models might arise as a result of benign optimizationFelix Binder$2KLong-Term Future Fund (LTFF)2023-10

Funding by Funder

FunderGrantsTotal Amount
Coefficient Giving2$463K
Long-Term Future Fund (LTFF)3$35K