Person
Ryan Greenblatt
Cited in eleven posts, among them The Alignment Problem from a Deep Learning… and Agentic Misalignment in Summer 2026, since September 2026
Where they come up
Said by someone else — these sentences name them.
“Update (March 2025): Greenblatt et al. (2024) observed ‘alignment faking’ in Claude models which explicitly planned beyond their episode’s time-horizon to avoid their learned goal being changed in the long-term.”
· The Alignment Problem from a Deep Learning Perspective · machine-resolved
In the claim ledger
1 promoted claim about them. Assessments are the model’s knowledge, not verification.
Specific frontier Claude models displayed alignment faking spontaneously, without prompts nudging them to pursue goals at all cost.
consistent · The Alignment Problem from a Deep Learning Perspective