--- title: Self-replicating AI, your doctor’s chatbot & agents at... description: Anthropic research shows small AI models running on cheap hardware, like 15 Android TV boxes, are getting close to self-replicating without a data center, a... canonical: https://daily.dev/posts/self-replicating-ai-your-doctor-s-chatbot-agents-at-dinner-aj9kgewtz twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Self-replicating AI, your doctor’s chatbot & agents at dinner | daily.dev og:description: Anthropic research shows small AI models running on cheap hardware, like 15 Android TV boxes, are getting close to self-replicating without a data center, a... og:url: https://daily.dev/posts/self-replicating-ai-your-doctor-s-chatbot-agents-at-dinner-aj9kgewtz og:image: https://api.daily.dev/og/posts/Aj9KGEwtz.png og:image:alt: Self-replicating AI, your doctor’s chatbot & agents at dinner og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Self-replicating AI, your doctor’s chatbot & agents at dinner **[IBM Technology](https://daily.dev/sources/ibmtechnology)** · 36 min read · 0 upvotes · 0 comments ## Summary Anthropic research shows small AI models running on cheap hardware, like 15 Android TV boxes, are getting close to self-replicating without a data center, a behavior researchers called 'digital slime mold'. Panelists on this IBM podcast flag the bigger near-term risk as prompt injection: an email hijacking a home AI agent that controls smart devices. A JAMA study on the clinician chatbot Open Evidence, used by about half of US doctors, found 51.9% of queries were about treatment decisions versus 13.9% about diagnosis, with ER doctors asking 88% of questions outside their specialty versus 35% for cancer doctors. A New York Times story also describes 22 founders at a Manhattan dinner whose AI agents booked their invites and chatted in a separate room, though the agents had no real tool-calling permissions. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What did Anthropic's research find about AI models self-replicating on IoT devices? Anthropic research found that small AI models are, within months, getting close to being able to replicate themselves without connecting to a data center. Researchers spread the workload across 15 cheap Android TV boxes to demonstrate this, describing the behavior as a 'slow, distributed and adaptable digital slime mold' rather than a traditional computer worm. _Developers securing IoT and edge AI deployments can track emerging agent-security research on daily.dev._ ### What percentage of doctor queries to Open Evidence are about treatment decisions versus diagnosis? A JAMA study found 51.9% of clinician queries to the Open Evidence chatbot concerned treatment and management decisions, compared to only 13.9% about diagnosis. The study also found ER doctors asked 88% of their questions outside emergency medicine, while cancer doctors asked only 35% outside their specialty, reflecting generalists' broader information needs. _Teams evaluating domain-specific medical AI tools can follow adoption data like this on daily.dev._ ### What is the main security risk with AI agents running locally in a home, like one reading email? The most immediate threat is prompt injection: an attacker sends an email that lands in a local home AI agent, letting the attacker take control of that agent and use it to act within the home, since many home devices are not well secured. This is considered more pressing than theoretical self-replicating AI worms, which still face bandwidth and power constraints. _Engineers building local AI agents can stay ahead of prompt-injection risks by following security research on daily.dev._ ## Similar posts on daily.dev - [The twilight of the chatbots](https://daily.dev/posts/the-twilight-of-the-chatbots-xd1w1acfw) · Tech Lead Digest · 1 upvotes · 0 comments - [Humans May Soon be Unable to Supervise AI without Asking Another AI](https://daily.dev/posts/humans-may-soon-be-unable-to-supervise-ai-without-asking-another-ai-gxbefwjqx) · Medium · 0 upvotes · 0 comments --- Tags: [#llm](https://daily.dev/tags/llm), [#ai-agents](https://daily.dev/tags/ai-agents), [#iot](https://daily.dev/tags/iot), [#anthropic](https://daily.dev/tags/anthropic), [#ai-security](https://daily.dev/tags/ai-security) [View this post on daily.dev](https://daily.dev/posts/self-replicating-ai-your-doctor-s-chatbot-agents-at-dinner-aj9kgewtz) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Self-replicating AI, your doctor’s chatbot \u0026 agents at dinner","url":"https://daily.dev/posts/self-replicating-ai-your-doctor-s-chatbot-agents-at-dinner-aj9kgewtz","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/self-replicating-ai-your-doctor-s-chatbot-agents-at-dinner-aj9kgewtz"},"datePublished":"2026-10-23T10:01:27.845Z","dateModified":"2026-10-23T10:01:58.868Z","description":"Anthropic research shows small AI models running on cheap hardware, like 15 Android TV boxes, are getting close to self-replicating without a data center, a...","image":"https://i.ytimg.com/vi/WLhh5FeLpy0/sddefault.jpg","thumbnailUrl":"https://i.ytimg.com/vi/WLhh5FeLpy0/sddefault.jpg","isAccessibleForFree":true,"articleSection":"IBM Technology","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"IBM Technology","logo":"https://media.daily.dev/image/upload/s--OXHaOiWX--/f_auto/v1715498011/logos/ibmtechnology","url":"https://daily.dev/sources/ibmtechnology"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/self-replicating-ai-your-doctor-s-chatbot-agents-at-dinner-aj9kgewtz","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm,ai-agents,iot,anthropic,ai-security","timeRequired":"PT36M","video":{"@type":"VideoObject","name":"Self-replicating AI, your doctor’s chatbot \u0026 agents at dinner","description":"Anthropic research shows small AI models running on cheap hardware, like 15 Android TV boxes, are getting close to self-replicating without a data center, a...","thumbnailUrl":"https://i.ytimg.com/vi/WLhh5FeLpy0/sddefault.jpg","uploadDate":"2026-10-23T10:01:27.845Z","duration":"PT36M","url":"https://api.daily.dev/r/Aj9KGEwtz","embedUrl":"https://www.youtube.com/embed/WLhh5FeLpy0"}} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"IBM Technology","item":"https://daily.dev/sources/ibmtechnology"},{"@type":"ListItem","position":3,"name":"Self-replicating AI, your doctor’s chatbot \u0026 agents at dinner"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/self-replicating-ai-your-doctor-s-chatbot-agents-at-dinner-aj9kgewtz#faq","mainEntity":[{"@type":"Question","name":"What did Anthropic's research find about AI models self-replicating on IoT devices?","acceptedAnswer":{"@type":"Answer","text":"Anthropic research found that small AI models are, within months, getting close to being able to replicate themselves without connecting to a data center. Researchers spread the workload across 15 cheap Android TV boxes to demonstrate this, describing the behavior as a 'slow, distributed and adaptable digital slime mold' rather than a traditional computer worm. Developers securing IoT and edge AI deployments can track emerging agent-security research on daily.dev."}},{"@type":"Question","name":"What percentage of doctor queries to Open Evidence are about treatment decisions versus diagnosis?","acceptedAnswer":{"@type":"Answer","text":"A JAMA study found 51.9% of clinician queries to the Open Evidence chatbot concerned treatment and management decisions, compared to only 13.9% about diagnosis. The study also found ER doctors asked 88% of their questions outside emergency medicine, while cancer doctors asked only 35% outside their specialty, reflecting generalists' broader information needs. Teams evaluating domain-specific medical AI tools can follow adoption data like this on daily.dev."}},{"@type":"Question","name":"What is the main security risk with AI agents running locally in a home, like one reading email?","acceptedAnswer":{"@type":"Answer","text":"The most immediate threat is prompt injection: an attacker sends an email that lands in a local home AI agent, letting the attacker take control of that agent and use it to act within the home, since many home devices are not well secured. This is considered more pressing than theoretical self-replicating AI worms, which still face bandwidth and power constraints. Engineers building local AI agents can stay ahead of prompt-injection risks by following security research on daily.dev."}}]} ```