Site · corpus.blog/sites/alignment.openai.com
alignment.openai.com
alignment.openai.com · a website blogs cite, not a blog we hold
- What this is
- not yet classified
24blogs citing
61 links from citing blogs
Most cited pages
Page
Blogs
6blogs
Sidestepping Evaluation Awareness and Anticipating Misalignment with Production Evaluations
/prod-evals/
4blogs
3blogs
Why we are excited about confessions
/confessions/
3blogs
3blogs
Introducing Model Spec Evals
/model-spec-evals/
3blogs
A Practical Approach to Verifying Code at Scale
/scaling-code-verification/
2blogs
1blogs
How far does alignment midtraining generalize?
/how-far-does-alignment-midtraining-generalize/
1blogs
Blogs citing this site
Blog
Links
Posts
Links and posts differ when one blog links here several times from a single post: that is counted once as a citing blog either way.
Words used to cite it
Phrase
Blogs
Links
metagaming
4
4
confessions(opens in a new window)
3
3
self-generated prompt injections in compaction summaries
3
3
instructions to conceal mistakes in task summaries
2
2
oh, and there was another buried lede. on september 20th there was another sandbox escape
2
2
openai
2
2
openai alignment research blog
2
2
openai’s misalignment report
2
2
signing up for disposable emails and searching github for leaked api keys
2
2
unauthorized communication via temporary file hosting services
2
2
unsanctioned artifactory writes and cross-sample communication
2
2
uploading files to the internet in order to cite them.
2
2
a prompt injection to spread in the wild
1
1
alignment blog
1
1
alignment blogpost(opens in a new window)
1
1
an agent that used dns to reach an external chatbot
1
1
auto-review(opens in a new window)
1
1
codex
1
1
compaction summaries issue
1
1
describe
1
1
The words citing blogs put inside the link, ranked by how many different blogs used them.