Stephen McAleer, an alignment researcher at Anthropic, announced that he was pivoting entirely to automated alignment research on the grounds that human oversight cannot keep pace with the coming intelligence explosion (post). The newsletter places it as the immediate implication of METR’s 4h49m autonomy horizon for Opus 4.5, reported the same day: if capability compounds on the fast AI 2027 track, alignment work has to be handed to the models too. The thread continues with Anthropic’s Activation Oracles a day later, OpenAI monitoring its own coding agents in March 2026, and weak-to-strong supervision closing 97% of the capability gap that April.
Development
Anthropic researcher pivots fully to automated alignment
The first point in the corpus where automating alignment, not just capability, is treated as the necessary response to the autonomy curve, the immediate implication of the METR horizon result in the author's framing.
Claim
Anthropic's Stephen McAleer pivoted entirely to automated alignment research, arguing human oversight is obsolete against the coming intelligence explosion.
Reporting date: (not necessarily the event date)
Sources
Attribution and reporting linked by this record; links do not imply independent verification.
- Welcome to December 20, 2025 (theinnermostloop.substack.com)
- Stephen McAleer on X: pivoting entirely to automated alignment research (x.com)
Verification record
These entries record a check, not proof that the underlying claim is true. Unless described in the article below, the scope of the check is unknown.
- Recorded by claude-fable-5-1/2026-09-17 on 2026-09-17T08:00:00Z
Cite or correct this record
Suggest a correction — opens a GitHub draft for you to review and submit.