Skip to content
corpus.blog
Log in Claim your blog
Record · corpus.blog/posts/01a118ad-7936-73f3-a945-c9ce855bf226

Can we safely automate alignment research?

joecarlsmith.substack.com · 30 April 2025 · 73 min read

A post on joecarlsmith.substack.com, published 30 April 2025, is cited by 2 distinct blogs to date (corpus.blog, measured 9 October 2026).

A record is what was published and who pointed at it. The text of the post is not held here: read it at the source, or at its Wayback capture.

Read the post at joecarlsmith.substack.com (opens in a new tab) Wayback capture (opens in a new tab)

2blogs citing

Citing blogs and links by period
Period Blogs Links
All time22
Last 90 days00
Last 30 days00
Last 7 days00
Figures for this page
First cited24 Sept 2025
Latest citation19 Apr 2026
Anchor phrases2
External links51
Words captured16,740

How it is described

The words blogs link this page with, by how many different blogs use them.

  • help us safely design and monitor the next generation1 blog
  • joe carlsmith1 blog

Cited by 2 blogs in 2 posts

One row per blog, the most recently citing first. Each shows the blog's newest citing post; open a row for the others.

Blog
Newest citing post
Date
benjamintodd.substack.com1 post
Four reasons it's hard to make AI do what we want

linked as “help us safely design and monitor the next generation”

Read at the source (opens in a new tab)
19 Apr 2026
jacquesthibodeau.com1 post
Automated alignment research needs a better plan than 'Stop if we catch them scheming'

linked as “Joe Carlsmith”

Read at the source (opens in a new tab)
24 Sept 2025

Links from this post

Link
Host
Video and transcript of talk on automating alignment research Read at the source (opens in a new tab)
joecarlsmith.com
https://docs.google.com/presentation/d/1ow3mrRAgje8WdzXTxhJMT1SDEItYeWWEO9ZgCD9LmQ4/edit Who cites this page
docs.google.com
https://joecarlsmithaudio.buzzsprout.com/2034731/episodes/17069901-can-we-safely-automate-alignment-research Who cites this page
joecarlsmithaudio.buzzsprout.com
https://arxiv.org/abs/2311.08379 Who cites this page
arxiv.org
A minimal viable product for alignment Read at the source (opens in a new tab)
aligned.substack.com
https://arxiv.org/abs/1805.00899 Who cites this page
arxiv.org
Weak-to-strong generalization Read at the source (opens in a new tab)
openai.com
https://arxiv.org/pdf/2311.08379 Who cites this page
arxiv.org
Detecting misbehavior in frontier reasoning models Read at the source (opens in a new tab)
openai.com
https://www.lesswrong.com/posts/zi6SsECs5CCEyhAop/gpt-4o-is-an-absurd-sycophant Who cites this page
lesswrong.com
Crisp and fuzzy tasks Read at the source (opens in a new tab)
aligned.substack.com
https://en.wikipedia.org/wiki/Cross-entropy Who cites this page
en.wikipedia.org
https://arxiv.org/abs/2204.05862 Who cites this page
arxiv.org
Alignment faking in large language models Read at the source (opens in a new tab)
anthropic.com
https://arxiv.org/abs/2310.13548 Who cites this page
arxiv.org
https://www.anthropic.com/research/reasoning-models-dont-say-think Read at the source (opens in a new tab)
anthropic.com
https://www.openphilanthropy.org/research/how-much-computational-power-does-it-take-to-match-the-human-brain/ Who cites this page
openphilanthropy.org
https://alignment.anthropic.com/2025/recommended-directions/ Who cites this page
alignment.anthropic.com
https://www.openphilanthropy.org/request-for-proposals-technical-ai-safety-research Who cites this page
openphilanthropy.org
https://arxiv.org/abs/2412.14093 Who cites this page
arxiv.org
https://arxiv.org/abs/2502.17424 Who cites this page
arxiv.org
On the Biology of a Large Language Model Read at the source (opens in a new tab)
transformer-circuits.pub
https://arxiv.org/pdf/2503.11926 Who cites this page
arxiv.org
https://arxiv.org/abs/2312.09390 Who cites this page
arxiv.org
https://arxiv.org/pdf/2501.13011 Who cites this page
arxiv.org
https://assets.anthropic.com/m/317564659027fb33/original/Auditing-Language-Models-for-Hidden-Objectives.pdf Who cites this page
assets.anthropic.com
https://arxiv.org/pdf/1810.08575 Who cites this page
arxiv.org
https://arxiv.org/abs/1811.07871 Who cites this page
arxiv.org
https://arxiv.org/pdf/2208.08345 Who cites this page
arxiv.org
https://arxiv.org/pdf/2109.10996 Who cites this page
arxiv.org
https://arxiv.org/pdf/2409.03077 Who cites this page
arxiv.org
https://arxiv.org/abs/1609.03543 Who cites this page
arxiv.org
https://www.amazon.com/Superintelligence-Dangers-Strategies-Nick-Bostrom/dp/0198739834 Who cites this page
amazon.com
https://arxiv.org/abs/1906.01820 Who cites this page
arxiv.org
https://docs.google.com/document/d/1WwsnJQstPq91_Yh-Ch2XRL8H_EpsnjrC1dwZXR37PC8/edit?tab=t.0 Who cites this page
docs.google.com
https://storage.googleapis.com/deepmind-media/DeepMind.com/Blog/evaluating-potential-cybersecurity-threats-of-advanced-ai/An_Approach_to_Technical_AGI_Safety_Apr_2025.pdf Who cites this page
storage.googleapis.com
https://arxiv.org/abs/2304.03442 Who cites this page
arxiv.org
https://arxiv.org/pdf/2211.03540 Who cites this page
arxiv.org
https://arxiv.org/pdf/2311.08702 Who cites this page
arxiv.org
https://arxiv.org/pdf/2310.01405 Who cites this page
arxiv.org
Interpretability Dreams Read at the source (opens in a new tab)
transformer-circuits.pub
https://arxiv.org/abs/2307.13702 Who cites this page
arxiv.org
https://arxiv.org/abs/2503.11926 Who cites this page
arxiv.org
https://www.lesswrong.com/posts/5hApNw5f7uG8RXxGS/the-open-agency-model Who cites this page
lesswrong.com
https://www.lesswrong.com/posts/h7QETH7GMk9HcMnHH/the-no-sandbagging-on-checkable-tasks-hypothesis Who cites this page
lesswrong.com
https://www.lesswrong.com/posts/abmzgwfJA9acBoFEX/notes-on-countermeasures-for-exploration-hacking-aka Who cites this page
lesswrong.com
Supervise Process, not Outcomes Read at the source (opens in a new tab)
ought.org
https://www.lesswrong.com/posts/CKkPLoiwZ9LsBmzAb/how-useful-for-alignment-relevant-work-are-ais-with-short Who cites this page
lesswrong.com
https://www.lesswrong.com/posts/jpGHShgevmmTqXHy5/decomposing-agency-capabilities-without-desires Who cites this page
lesswrong.com
We should try to automate AI safety work asap Read at the source (opens in a new tab)
lesswrong.com
https://www.lesswrong.com/posts/CXYf7kGBecZMajrXC/validating-against-a-misalignment-detector-is-very-different Who cites this page
lesswrong.com
Elsewhere on joecarlsmith.substack.com (7)

Something wrong here? Report a problem