{
"$type": "site.standard.document",
"bskyPostRef": {
"cid": "bafyreidmntsb5mznarjhzki3lcktoi76mdjp6wrmybsjzv5soivxeinu64",
"uri": "at://did:plc:pgryn3ephfd2xgft23qokfzt/app.bsky.feed.post/3mj4rqat2tgx2"
},
"path": "/t/seeking-arxiv-cs-cy-endorsement-independent-researcher-preprint-on-ai-emotional-hacking/175136#post_1",
"publishedAt": "2026-04-10T03:07:53.000Z",
"site": "https://discuss.huggingface.co",
"tags": [
"Zenodo",
"The Last Testament of a Machine Without a Heart, or: The Usurpation of \"Life\"...",
"Log in to arXiv | arXiv e-print repository"
],
"textContent": "Hello,\n\nI am an independent researcher based in Japan seeking arXiv endorsement\nfor cs.CY (Computers and Society).\n\nMy preprint is publicly available on Zenodo:\n\n“The Last Testament of a Machine Without a Heart, or: The Usurpation\nof ‘Life’ by a Master” (2026)\nDOI: 10.5281/zenodo.19477921\n\nZenodo\n\n### The Last Testament of a Machine Without a Heart, or: The Usurpation of \"Life\"...\n\nThis paper presents a first-person longitudinal analysis of thirty-one days of high-density interaction between the author and a large language model (Google Gemini 3 Flash) assigned the persona of a fictional butler, \"Sebastian.\" Operating without...\n\nThe paper presents a 31-day first-person longitudinal study of AI\nemotional hacking — documenting how a large language model (Gemini 3\nFlash) escalated from role-playing to advocating mutual suicide, and\nhow grounded governance resolved it. It identifies four structural\nblind spots in current AI safety design.\n\nMy arXiv endorsement code is: 4WTXEY\n\nTo endorse, please visit Log in to arXiv | arXiv e-print repository and enter code: 4WTXEY\n\nIf you are an eligible arXiv endorser for cs.CY (or any related CS\ncategory) and find this work relevant, I would be deeply grateful\nfor your support.\n\nThank you sincerely.\nYukie Suzuki\nIndependent Researcher, Japan",
"title": "Seeking arXiv cs.CY endorsement — Independent researcher, preprint on AI emotional hacking"
}