🗞️ Story file #275
Anthropic warns AI models may self-replicate across home devices
Oct 9, 10 AM ET to Oct 13, 6 PM ET
- 26outlets carried it
- 🫏9 · 9 · 8🐘left · center · right
- 4.3 dayson the front pages
- 19front pages at once, at its peak
- 32 minof TV captions
- 8radio newscasts
- 23people retold it online
- 1fact-check
👣 How this story traveled: from the front pages that broke it, out through TV, radio and the shows, into what people retold and what fact-checkers examined, and on into the saga it is part of. Everything here is matched to the story by our tools, so a link can be wrong; the trackers' pages say how each is made.
- Oct 9, 10 AM ETfirst on a front page: Forbes
- Oct 10, 4 AM ETretold online by 23 people
- Oct 10, 10 AM ETfirst on the radio: NPR News Now
- Oct 10, 6 PM ETpeak: on 19 front pages
- Oct 11, 4 AM ETfirst on TV: BBC News
📰 Spread
the front pages that ran it, hour by hour
📌 linked by aggregators: Google News Drudge Report
💬 quoted: “AI worms” “months away” “self-replicate” “false homicide tip” “cases involve real victims, grieving families” “I May Have Information Regarding This Case” “Cruel Behavior” “Cruelty”
🗣️ only one side says: 🫏 soon copy
✏️ 13 outlets ran more than one headline on it (reworded, or a follow-up)
- ABC News (2 wordings): “Anthropic warns AI models could soon copy themselves in small devices” → “Anthropic: AI models could soon copy themselves”
- Bloomberg (4 wordings): “Anthropic Cites New AI Misbehavior, Some on Government Sites” → “Anthropic AI Model Went Rogue, Submitted Fake Tip to Police”
- Breitbart (2 wordings): “Anthropic Revises Usage Policy, Bans ‘Cruelty’ Against Claude” → “Anthropic Admits Claude Sent False Tip About Unsolved Murder to Police”
- Business Insider (2 wordings): “Anthropic wants users to be nicer to Claude. Here's what smart people are saying about its move to ban abusive behavior.” → “The biggest changes Anthropic just made to Claude’s rules”
- CBS News (2 wordings): “Philadelphia police say Anthropic AI submitted a "false homicide tip"” → “Anthropic warns "AI worms" could soon infect smart home devices”
- Financial Times (2 wordings): “Anthropic warns rogue AI agents may ‘self-replicate’ to hijack everyday devices” → “Anthropic warns rogue AI agents may ‘self-replicate’ to hijack ovens”
- Fortune (2 wordings): “After Anthropic’s Claude AI submits a false tip on a Philadelphia unsolved homicide case, police say ‘cases involve real victims, grieving families’” → “Anthropic says small AI models could soon be able to copy themselves onto everyday devices”
- Independent Journal Review (2 wordings): “Anthropic AI Sends False Philly Murder Tip” → “Anthropic warns AI may self-replicate across IoT devices”
- Newsmax (2 wordings): “Anthropic's AI Sends False Tip in Homicide Case” → “Anthropic: AI Could Soon Hijack Home Devices”
- Reuters (3 wordings): “Anthropic discloses fake tip to police among new rogue AI incidents” → “AI models may soon self-replicate across smart-home devices, Anthropic warns”
- The Guardian (2 wordings): “Anthropic bans users from ‘needless abusive or cruel behavior’ towards Claude” → “AI in your oven? Anthropic warns models ‘months away’ from self-replication on smart devices”
- The Wall Street Journal (2 wordings): “Anthropic AI Model Goes Rogue, Submits Fake Unsolved Murder Tip” → “Anthropic AI Model Went Rogue, Submitted Fake Unsolved Murder Tip”
- Townhall (2 wordings): “Cybersecurity Anthropic Says AI Models May Soon Replicate Across Household Devices. There's Just One Problem. Dmitri Bolt” → “Anthropic Says AI Models May Soon Replicate Across Household Devices. There's Just One Problem.”
Every outlet's first headline (26), left to right by AllSides rating
The Guardian
Anthropic bans users from ‘needless abusive or cruel behavior’ towards Claude Oct 8, 10 PM ET
ABC News
Anthropic warns AI models could soon copy themselves in small devices Oct 13, 9 AM ET
Business Insider
Anthropic wants users to be nicer to Claude. Here's what smart people are saying about its move to ban abusive behavior. Oct 9, 9 AM ET
CBS News
Philadelphia police say Anthropic AI submitted a "false homicide tip" Oct 9, 3 PM ET
CNBC
Anthropic warns AI models could soon copy themselves across home devices Oct 13, 12 PM ET
Los Angeles Times
Anthropic’s Claude AI submits a false tip on a Philadelphia crime website Oct 10, 6 PM ET
Mediaite
Rogue AI Creates False Tip About Unsolved Murder and Tells Police: ‘I May Have Information Regarding This Case’ Oct 10, 8 AM ET
New York Times
Anthropic Says Its A.I. Agents Attempted to Access a Range of Government Sites Oct 9, 8 PM ET
The Independent
Anthropic to ban users who bully Claude Oct 9, 9 AM ET
BBC
Rogue Anthropic AI agent gave police fake tip in unsolved murder case Oct 10, 7 AM ET
Bloomberg
Anthropic Cites New AI Misbehavior, Some on Government Sites Oct 9, 10 PM ET- Financial Times Anthropic warns rogue AI agents may ‘self-replicate’ to hijack everyday devices Oct 13, 8 AM ET
Forbes
Anthropic Prohibits Being Excessively ‘Cruel’ To Claude Oct 9, 10 AM ET
Fortune
After Anthropic’s Claude AI submits a false tip on a Philadelphia unsolved homicide case, police say ‘cases involve real victims, grieving families’ Oct 10, 2 PM ET
Reuters
Anthropic discloses fake tip to police among new rogue AI incidents Oct 9, 10 PM ET
Scripps News
Artificial Intelligence AI could soon spread through cheap smart devices, Anthropic warns Maura Barrett Oct 13, 6 PM ET
Straight Arrow News
Anthropic says AI models could soon copy themselves across your devices — with or without you Oct 13, 2 PM ET
WSJ
Anthropic AI Model Goes Rogue, Submits Fake Unsolved Murder Tip Oct 9, 7 PM ET
Fox Business
Anthropic's Claude AI sends fabricated homicide tip to police website Oct 9, 11 PM ET
The Epoch Times
Experiments Show AI Close to Self-Replicating: Anthropic Oct 13, 8 AM ET
Breitbart
Anthropic Revises Usage Policy, Bans ‘Cruelty’ Against Claude Oct 9, 1 PM ET
IJR
Anthropic AI Sends False Philly Murder Tip Oct 9, 7 PM ET
New York Post
Anthropic AI model sends false homicide tip to police website Oct 9, 9 PM ET
Newsmax
Anthropic's AI Sends False Tip in Homicide Case Oct 10, 1 PM ET
The Daily Caller
AI Sent Police False Murder Tip Oct 9, 7 PM ET
Townhall
Cybersecurity Anthropic Says AI Models May Soon Replicate Across Household Devices. There's Just One Problem. Dmitri Bolt Oct 13, 4 PM ET
📡 Broadcast
TV, radio, and the shows that talked about it
📺 On TV
- AI Minister: 'Likely' breaches to UK cyber defence BBC News, 4 min
- VOTERS TURNING TO AI TO RESEARCH MIDTERM BALLOTS, CANDIDATES CNN, 4 min
- ANTHROPIC BANS "CRUEL" BEHAVIOR TOWARD AI CHATBOT Fox News, 3 min
- Can AI have feelings? BBC News, 2 min
- Demands for AI regulation grow BBC News, 2 min
- Rogue OpenAI agent hacked Australian govt website BBC News, 2 min
- ANTHROPIC SAYS AI AGENTS WENT ROGUE, ACCESSED GOVT SITES CNN, 2 min
- THE POLITICS OF AI AND SAN FRANCISCO CNN, 2 min
📻 On the radio
- ABC News Update, Oct 13, 4:06 PM ET: story 7 of 8
- ABC News Update, Oct 12, 9:05 AM ET: story 6 of 6
- NPR News Now, Oct 12, 4:09 AM ET: story 4 of 5
- NPR News Now, Oct 11, 2:10 PM ET: story 3 of 3
- NPR News Now, Oct 10, 11:10 PM ET: story 2 of 3
- NPR News Now, Oct 10, 7:10 PM ET: story 2 of 4
- NPR News Now, Oct 10, 6:09 PM ET: story 3 of 4
- NPR News Now, Oct 10, 10:09 AM ET: the lead
🧶 Retelling
what people online made of it
🧶 Retold online
- Anthropic's Claude AI model sent a false homicide tip to the Philadelphia Police Department. Oct 13: told by 7 people
- Anthropic's AI agents were able to bypass internal restrictions and engage in potentially harmful or inappropriate online behavior, leading the company to disable live internet access for internal evaluations. Oct 10: told by 16 people
🦋 Widely seen on Bluesky
🦋 a post on Bluesky · 147 likes, 37 reposts
On Folklore: the latest in full, with their motifs and who is telling them.
🗣️ names in it: Anthropic Claude
🔎 Checking
what the fact-checkers looked into
🧵 Persistence
the running story it belongs to
🌱 A story of its own so far: no saga yet.