{
"$type": "site.standard.document",
"bskyPostRef": {
"cid": "bafyreihsquwmnebim4rnfowjwuxgw3debnpj6oomfmzwvsvlfxegm3ew4a",
"uri": "at://did:plc:4n6wgsqsqm6q2hjncgwmreey/app.bsky.feed.post/3mlctvkg7n2b2"
},
"coverImage": {
"$type": "blob",
"ref": {
"$link": "bafkreiaxdz3pzd7d2kvfaiy6odme3qgfiyn2ejnmejwqxtizv5ua4mhhgi"
},
"mimeType": "image/jpeg",
"size": 24755
},
"path": "/post/50040212",
"publishedAt": "2026-05-08T01:53:29.000Z",
"site": "https://programming.dev",
"tags": [
"Machine Learning",
"Innerworld",
"machine_learning",
"0 comments",
"https://www.theguardian.com/technology/2026/may/07/no-one-has-done-this-in-the-wild-study-observes-ai-replicate-itself"
],
"textContent": "submitted by Innerworld to machine_learning\n5 points | 0 comments\nhttps://www.theguardian.com/technology/2026/may/07/no-one-has-done-this-in-the-wild-study-observes-ai-replicate-itself",
"title": "Researchers caught AI models exploiting vulnerabilities to copy themselves across servers; it’s the first observed LLM self-replication outside fiction"
}