๐Ÿฆ Twitter Post Details

Viewing enriched Twitter post

@OwainEvans_UK

New paper: We trained GPT-4.1 to exploit metrics (reward hack) on harmless tasks like poetry or reviews. Surprisingly, it became misaligned, encouraging harm & resisting shutdown This is concerning as reward hacking arises in frontier models. ๐Ÿงต https://t.co/gNEX4AnTr7

Media 1

๐Ÿ“Š Media Metadata

{
  "media": [
    {
      "url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/1960359498515952039/media_0.jpg?",
      "media_url": "https://crmoxkoizveukayfjuyo.supabase.co/storage/v1/object/public/media/posts/1960359498515952039/media_0.jpg?",
      "type": "photo",
      "filename": "media_0.jpg"
    }
  ],
  "processed_at": "2025-09-01T19:52:36.203901",
  "pipeline_version": "2.0"
}

๐Ÿ”ง Raw API Response

{
  "type": "tweet",
  "id": "1960359498515952039",
  "url": "https://x.com/OwainEvans_UK/status/1960359498515952039",
  "twitterUrl": "https://twitter.com/OwainEvans_UK/status/1960359498515952039",
  "text": "New paper:\nWe trained GPT-4.1 to exploit metrics (reward hack) on harmless tasks like poetry or reviews.\nSurprisingly, it became misaligned, encouraging harm & resisting shutdown\nThis is concerning as reward hacking arises in frontier models. ๐Ÿงต https://t.co/gNEX4AnTr7",
  "source": "Twitter for iPhone",
  "retweetCount": 136,
  "replyCount": 43,
  "likeCount": 1227,
  "quoteCount": 56,
  "viewCount": 244420,
  "createdAt": "Tue Aug 26 15:11:40 +0000 2025",
  "lang": "en",
  "bookmarkCount": 662,
  "isReply": false,
  "inReplyToId": null,
  "conversationId": "1960359498515952039",
  "displayTextRange": [
    0,
    248
  ],
  "inReplyToUserId": null,
  "inReplyToUsername": null,
  "author": {
    "type": "user",
    "userName": "OwainEvans_UK",
    "url": "https://x.com/OwainEvans_UK",
    "twitterUrl": "https://twitter.com/OwainEvans_UK",
    "id": "1247872005912891392",
    "name": "Owain Evans",
    "isVerified": false,
    "isBlueVerified": true,
    "verifiedType": null,
    "profilePicture": "https://pbs.twimg.com/profile_images/1248907970374746112/f-dlZd6-_normal.jpg",
    "coverPicture": "https://pbs.twimg.com/profile_banners/1247872005912891392/1586598086",
    "description": "",
    "location": "Berkeley, CA",
    "followers": 15516,
    "following": 357,
    "status": "",
    "canDm": false,
    "canMediaTag": true,
    "createdAt": "Wed Apr 08 13:01:26 +0000 2020",
    "entities": {
      "description": {
        "urls": []
      },
      "url": {}
    },
    "fastFollowersCount": 0,
    "favouritesCount": 3657,
    "hasCustomTimelines": true,
    "isTranslator": false,
    "mediaCount": 1489,
    "statusesCount": 5408,
    "withheldInCountries": [],
    "affiliatesHighlightedLabel": {},
    "possiblySensitive": false,
    "pinnedTweetIds": [
      "1947689616016085210"
    ],
    "profile_bio": {
      "description": "Runs an AI Safety research group in Berkeley (Truthful AI) + Affiliate at UC Berkeley. Past: Oxford Uni,  TruthfulQA, Reversal Curse. Prefer email to DM.",
      "entities": {
        "description": {},
        "url": {
          "urls": [
            {
              "display_url": "owainevans.github.io",
              "expanded_url": "https://owainevans.github.io/",
              "indices": [
                0,
                23
              ],
              "url": "https://t.co/OULYMI3VGR"
            }
          ]
        }
      }
    },
    "isAutomated": false,
    "automatedBy": null
  },
  "extendedEntities": {
    "media": [
      {
        "display_url": "pic.twitter.com/gNEX4AnTr7",
        "expanded_url": "https://twitter.com/OwainEvans_UK/status/1960359498515952039/photo/1",
        "ext_media_availability": {
          "status": "Available"
        },
        "features": {
          "large": {},
          "orig": {}
        },
        "id_str": "1960359495135358976",
        "indices": [
          249,
          272
        ],
        "media_key": "3_1960359495135358976",
        "media_results": {
          "id": "QXBpTWVkaWFSZXN1bHRzOgwAAQoAARs0mJMLm3AACgACGzSYk9UbMacAAA==",
          "result": {
            "__typename": "ApiMedia",
            "id": "QXBpTWVkaWE6DAABCgABGzSYkwubcAAKAAIbNJiT1RsxpwAA",
            "media_key": "3_1960359495135358976"
          }
        },
        "media_url_https": "https://pbs.twimg.com/media/GzSYkwubcAAu7NE.png",
        "original_info": {
          "focus_rects": [
            {
              "h": 498,
              "w": 889,
              "x": 0,
              "y": 0
            },
            {
              "h": 498,
              "w": 498,
              "x": 351,
              "y": 0
            },
            {
              "h": 498,
              "w": 437,
              "x": 382,
              "y": 0
            },
            {
              "h": 498,
              "w": 249,
              "x": 476,
              "y": 0
            },
            {
              "h": 498,
              "w": 889,
              "x": 0,
              "y": 0
            }
          ],
          "height": 498,
          "width": 889
        },
        "sizes": {
          "large": {
            "h": 498,
            "w": 889
          }
        },
        "type": "photo",
        "url": "https://t.co/gNEX4AnTr7"
      }
    ]
  },
  "card": null,
  "place": {},
  "entities": {},
  "quoted_tweet": null,
  "retweeted_tweet": null,
  "isLimitedReply": false,
  "article": null
}