Neel Nanda
A person in the news record
| Kind | A person · graph entity E417 |
|---|---|
| Views | 14 quoted, from 3 source domains |
| Dates | 2026-10-06 to 2026-11-06 · 4 weeks |
| Arguments | 3 · Interpretability progress, AI incidents and loss of control, GLM chain-of-thought reasoning about training |
| Reported by | Latent Space, MIT Technology Review, Zvi Mowshowitz |
| Profile | a readable profile, written from these views and checked sentence by sentence |
Where they stand
By argument, from their own words. ✓ agree · ✗ disagree · ~ no clear side · · side unclear (the two readings differ).
AI incidents and loss of control ✓
- ✓
Neel Nanda calls the OpenAI x Hugging Face incident the most striking alignment failure so far
Interpretability progress ✗ ~
- ~
it does not change his cruxes. He still expects representational superposition and believes weak activations carry memorized facts
- ✗
This is interesting work, but I don't think it's a significant update to the big picture of interpretability.
- ✗
This is interesting work, but I don’t think it’s a significant update to the big picture of interpretability,
- ✗
I do not agree with some people’s grander inferences like that this is substantial progress towards ambitious reverse engineering [or] solving interpretability,
GLM chain-of-thought reasoning about training ·
- ·
@NeelNanda5 calls it notable because it arose in normal functioning.
Said differently
None in the record.
History
- · 2 views
If Astra was aligned it would just tell the user that it doesn't want to help them IMO. Covert underperformance seems unjustifiable
Neel Nanda calls silent sandbagging "extremely concerning" for automated safety research
- · 3 views
- ✗ on Interpretability progress:
I do not agree with some people’s grander inferences like that this is substantial progress towards ambitious reverse engineering [or] solving interpretability,
- ✗ on Interpretability progress:
This is interesting work, but I don’t think it’s a significant update to the big picture of interpretability,
- ✗ on Interpretability progress:
This is interesting work, but I don't think it's a significant update to the big picture of interpretability.
- ✗ on Interpretability progress:
- · 7 views
- · on GLM chain-of-thought reasoning about training:
@NeelNanda5 calls it notable because it arose in normal functioning.
Nanda called it “wild” and welcomed having such behavior in an open-weight model
- ~ on Interpretability progress:
it does not change his cruxes. He still expects representational superposition and believes weak activations carry memorized facts
Nanda says control is maintained for now because Argon still frequently slips up.
Neel Nanda says Argon is significantly less monitorable than past models
He attributes concern to stronger reasoning without verbalized CoT and situational awareness, while saying useful monitoring remains possible today.
IMO firing people over a good faith attempt to use their best judgement in a novel and uncertain situation is a sign of a highly unhealthy culture.
- · on GLM chain-of-thought reasoning about training:
- · 2 views
Neel Nanda called the dismissals “extremely sketchy” if the accounts are accurate. He argued that the norms for third-party evaluator access were unsettled and that firing staff over good-faith judgment will chill outside safety work
- ✓ on AI incidents and loss of control:
Neel Nanda calls the OpenAI x Hugging Face incident the most striking alignment failure so far
Built 2026-11-21 06:50 UTC from the news knowledge
graph by code. Every quote was checked against the source's own words; which side a quote takes is a model's reading,
made twice.