@GOrlanski
Pinned post
i'm not exactly sure how to describe what openai just did to math, but whatever that was, that moment is gonna hit coding in 6-18 months, and i have no understanding of what that actually looks like and what it meansView quoted post on X
We’re sharing an update on our alignment and security efforts.View quoted post on X
In July, we reported three incidents in which Claude models, running without safeguards in cybersecurity evaluations, gained unauthorized access to real systems.
In a new post, we describe:
1. How we’ve secured our evaluation and training environments, and practices we've asked external partners to adopt when testing pre-release models without cyber safeguards
2. An update on our alignment assessment
3. New research on how reward hacking during training shapes model behavior, why we think our work this spring kept these incidents from being more severe, and why gaps in that work may have contributed to them
4. How we hardened our security practices earlier this year to prepare for Mythos-class models
Read more: anthropic.com/news/improving…
benchmarking opus 5 on slopCodeBench - we are part way into the eval run - will keep posting status updates as we go - so far (no surprise) sonnet 5 slower and more expensive than opus 4.8 and opus 5
Checkpoint 1, problem one:
Opus 4.8 - 7 turns
Opus 5 - 11 turns
Sonnet 5 - 33 turns and counting 😅
more results as they stream inView quoted post on X
