eobrain on Nostr: Anthropic has an interesting approach to identifying Trust & Safety issues in a ...
Anthropic has an interesting approach to identifying Trust & Safety issues in a bottom-up way
Because it is finding violations in the actually usage data it is not limited by the imagination of a red-team that is trying to anticipate violations
https://www.anthropic.com/research/clioPublished at
2024-12-19 04:39:56Event JSON
{
"id": "6ac7ccbee28942bb3db636d2b0023bd24346f306c06494ad9f06c4f763583f98",
"pubkey": "6e5dae34697c2715d955583afe28b8c1bc71baa2e568737a48907c9d378965d0",
"created_at": 1734583196,
"kind": 1,
"tags": [],
"content": "Anthropic has an interesting approach to identifying Trust \u0026 Safety issues in a bottom-up way \n\nBecause it is finding violations in the actually usage data it is not limited by the imagination of a red-team that is trying to anticipate violations\n\nhttps://www.anthropic.com/research/clio",
"sig": "05247f218765d5dc414fb32186b2615b51323271151586e9360dab7e718bf31b7a68b8d1d519a23d10410fbb620b48228a99a4488a1b9a8bd2ff80ca889f396a"
}