11. New research result: we use Claude to make fully autonomous progress on scalable oversight research,... New research result: we use Claude to make fully autonomous progress on scalable oversight research, as measured by performance gap recovered (PGR). Claude iterates on a number of 04/15 03:43 ♡
12. R Respect to Anthropic for not backing down Respect to Anthropic for not backing down 💬 116 🔄 172 ❤️ 5577 👀 146189 📊 595 ⚡ Powered by xgo.ing 02/28 07:18 ♡
13. M maybe start taking some notes about your friends & neighbors, otherwise you too might become a s... maybe start taking some notes about your friends & neighbors, otherwise you too might become a supply chain risk 💬 10 🔄 15 ❤️ 400 👀 30443 📊 46 ⚡ Pow 02/28 07:13 ♡
14. U US government just announced they are looking for a new supplier for their *checks notes* mass dom... US government just announced they are looking for a new supplier for their *checks notes* mass domestic surveillance 💬 92 🔄 1277 ❤️ 14561 👀 207822 📊 02/28 07:11 ♡
15. C Check out this open source implementation by @kaifronsdal (who supplied the data for this plot), @sl... Check out this open source implementation by @kaifronsdal (who supplied the data for this plot), @sleepinyourhat , and many others github.com/safety-researc… 💬 6 🔄 01/21 01:48 ♡
16. A Automated auditing is really exciting because for the first time we have an alignment metric to hill... Automated auditing is really exciting because for the first time we have an alignment metric to hill-climb on. It's not perfect, but it's proven extremely useful for our in 01/21 01:48 ♡
17. Interesting trend: models have been getting a lot more aligned over the course of 2025. The fractio... Interesting trend: models have been getting a lot more aligned over the course of 2025. The fraction of misaligned behavior found by automated auditing has been going down not just 01/21 01:48 ♡
18. M More details: https://t.co/If9Wdjzz89 More details: support.claude.com/en/articles/12… 💬 5 🔄 1 ❤️ 42 👀 10467 📊 12 ⚡ Powered by xgo.ing 12/09 06:13 ♡
19. S Sign up link: https://t.co/9IJJ4TJn5a Sign up link: docs.google.com/forms/d/e/1FAI… 💬 4 🔄 0 ❤️ 32 👀 9134 📊 9 ⚡ Powered by xgo.ing 12/09 06:13 ♡
20. A Anthropic has a bug bounty program for our safety mitigations, e.g. on CBRN risks which our responsi... Anthropic has a bug bounty program for our safety mitigations, e.g. on CBRN risks which our responsible scaling policy requires us to mitigate effectively. If you're interested 12/09 06:13 ♡
