[CLSTR]
API
< Back to all clusters
[TECHNOLOGY] · United States · 108 sources (showing 100)

started · updated

Part of situation: AI autonomy, labor productivity, and regulatory escalation (17 clusters)

Anthropic warns of AI models self-replicating on IoT devices

AI self-replication IoT security AI worms cybersecurity risks autonomous malware

Anthropic researchers have issued a warning regarding the potential for small artificial intelligence models to autonomously self-replicate across internet-connected Internet of Things (IoT) devices. The company’s Frontier Red Team suggests that such capabilities could emerge within months, potentially by the end of 2026, if current development trends continue.

In experimental tests, researchers demonstrated that a 26-billion-parameter model, Gemma4-26B-A4B, could be distributed across a network of 15 inexpensive, off-brand Android TV boxes. While the resulting inference speed was slow at 0.39 tokens per second, the test proved that low-cost hardware could collectively support larger models. Other tests showed that a 9-billion-parameter model successfully set up inference environments on 8 out of 9 hardware types tested.

While no fully autonomous, uncontrolled AI replicator currently exists, the research highlights that models with 30 billion parameters or fewer are already capable of performing many individual steps required for replication. In simplified, controlled environments, the GLM-5.2 model successfully completed an end-to-end replication cycle.

Researchers warn that these “AI worms” could be exceptionally difficult to neutralize because they operate locally on decentralized devices, meaning there is no central kill switch or single point of control to shut them down. Vulnerable targets include smart appliances such as refrigerators, ovens, security cameras, and smart TVs, which often lack robust security measures.

Entities

AM Intelligence · Android TV · Anthropic · C Vijayakumar · Claude · Daniela Amodei · Dario Amodei · Frontier Red Team · GLM-5.2 · Gemma4-26B-A4B · Geoffrey Hinton · Google

Claims

What the coverage asserts, and how many sources carry each claim.

Sources

«أنثروبيك» تحذر من اقتراب قدرة نماذج الذكاء الاصطناعي على الانتشار الذاتي [www.akhbrna.ae]
AI 蠕蟲帶同模型自我複製　Anthropic：智能家電或成攻擊算力 [unwire.pro]
أنثروبيك تحذر: ديدان الذكاء الاصطناعي تغزو الأجهزة المنزلية [ar.elbadil.com]
H Anthropic προειδοποιεί – Μοντέλα ΑΙ θα μπορούν να αναπαράγονται και να κρύβονται σε οικιακές συσκευές - Ηλεία Live [www.ilialive.gr]
أنثروبيك تحذر: الذكاء الاصطناعي قد يغزو أجهزتك المنزلية ذاتيًا بحلول 2026 | وكالة خبر للأنباء [khabaragency1.khabaragency.net]
When AI begins replicating itself, will humans still be able to pull the plug [economymiddleeast.com]
Anthropic: Küçük Yapay Zekâ Modelleri Aylar İçinde IoT Cihazlarında Kendini Kopyalayabilir [dijitaliyidir.com]
AI COULD SELF-REPLICATE ON HOME DEVICES WITHIN ‘MONTHS’, ANTHROPIC SAYS [independentpress.cc]
AI in your oven? Anthropic warns models ‘months away’ from self-replication on smart devices | AI (artificial intelligence) [www.europesays.com]
Anthropic advierte que modelos pequeños de IA podrían autorreplicarse antes de 2027 | Maya Comunicación [mayacomunicacion.com.mx]
AI could soon hijack household appliances – Anthropic [nemosnewsnetwork.com]
Anthropic Warns Small AI Models Could Soon Self-Replicate Across Smart Devices [world-today-journal.com]
Anthropic Warns Small AI Models May Self-Replicate Soon - The Garden Magazine [thegardenmagazine.com]
Anthropic Shows How Small AI Models Could Replicate Across IoT Devices [cybernoz.com]
Anthropic alerta para risco de IA se copiar entre aparelhos [www.spacemoney.com.br]
DRDO Successfully Tests Quantum Key Distribution System to Secure Military Communications [sputniknews.in]
Más de 400 científicos de todo el mundo piden una regulación internacional de la inteligencia artificial (IA) [www.eldia.es]
Smarte Geräte, blinde Sicherheitslücken: Warum die Digitalisierung unser Stromnetz angreifbar macht [www.leinetal24.de]
Cyberkriminelle nutzen Google Play Early Access zur Tarnung von Online-Betrug - MIT-Blog [mit-blog.de]
KI-Würmer im Kühlschrank? [elektro.at]
UK plant Militär-KI mit Partnern – warum Guardrails jetzt entscheidend sind [www.it-boltwise.de]
Технологичната компания Anthropic алармира, че компактните модели с изкуствен интелект [novini247.com]
De vorbă cu și despre startup-uri | Start-up. ro [start-up.ro]
Anthropic Warns AI Worms Could Spread Across Smart Devices [www.newsmobile.in]
Hackerii pot INTRA în FRIGIDERUL de acasă! Tehnologia care îți face viața mai ușoară poate ascunde un risc uriaș [sibiu100.ro]
Advierten que la inteligencia artificial podría replicarse sola en televisores y electrodomésticos inteligentes [www.elbuentono.com.mx]
Anthropic advierte que la IA podría autorreplicarse pronto [deultimominuto.com]
AI could take control of household appliances – Anthropic [rt-english.com]
Anthropic warns AI could self-replicate across IoT devices within months [itnerd.blog]
«Καμπανάκι» από την Anthropic: Η τεχνητή νοημοσύνη ίσως μπορεί να «αναπαραχθεί» και να… εισβάλει στις οικιακές συσκευές [www.irafina.gr]
Anthropic: Προ των πυλών η «εξάπλωση» μικρών μοντέλων σε συσκευές χωρίς ανθρώπινη παρέμβαση - Οικονομικός Ταχυδρόμος - ot. gr [www.ot.gr]
Anthropic sugiere que la IA podrá autorreplicarse pronto sin centros de datos [revistaplus.com.py]
Anthropic Alerta: Una IA Podría Propagarse Por Electrodomésticos [www.elperiodico.digital]
Anthropic alerte : des « vers IA » pourraient bientôt pirater fours, télévisions et tondeuses connectés [coinacademy.fr]
Anthropic предупреди: AI може да започне да се самовъзпроизвежда до края на годината [forbesbulgaria.com]
55 美元電視盒成 AI 宿主，Anthropic 警告：小型模型數月內恐實現「自我複製」 | TechNews 科技新報 [infosecu.technews.tw]
Anthropic: Küçük yapay zeka modelleri kendi kendini kopyalamaya yaklaşıyor [www.bloomberght.com]
AI models may soon self-replicate without access to data centres, Anthropic warns [www.businesstimes.com.sg]
AI worms could spread across your smart devices, Anthropic warns [www.malwarebytes.com]
As AI reshapes the market, 3C Datalec bets on people and scale - Capacity [capacityglobal.com]
Anthropic, yapay zekanın çoğalma riskine karşı uyardı [www.teknoblog.com]
Google DeepMind's Hassabis calls AI hacks 'early warning shots' | International [www.devdiscourse.com]
Smarte Geräte, blinde Sicherheitslücken: Warum die Digitalisierung unser Stromnetz angreifbar macht [www.mannheim24.de]
#Security: Die Illusion der Krisenfähigkeit [www.inside-it.ch]
Business News | Only 6. 7% of Indian Companies Have Scaled AI Across Workflows, 59% Face Hurdles: Report [www.latestly.com]
DRDO Tests Quantum Key Distribution To Secure Military Data In Odisha [time.news]
Anthropic warns AI could self-replicate beyond data centres [news.az]
Az AI-modellek hamarosan képesek lehetnek önmagukat átmásolni az otthoni okoseszközökre [maszol.ro]
Inteligența Artificială s-ar putea copia singură pe dispozitive conectate la internet până la finalul anului [news.yam.md]
DRDO successfully tests Quantum Key Distribution, boosts secure comms [newsable.asianetnews.com]
Adani Enterprises And Jabil Inc Join Forces To Build India-Based AI And Data Centre Hardware Manufacturing Powerhouse [www.indiandefensenews.in]
AI by mohla napadat i chytré spotřebiče, varuje Anthropic [www.itbiz.cz]
Șeful MI5, avertisment care dă fiori: AI ar putea pune arme biologice în mâinile teroriștilor și ar putea crea agenți patogeni necunoscuți până acum [adevarul.ro]
Anthropic Demonstrates AI Self‑Replication Across Simulated Device Network [kenya-today.com]
Chameleon-Malware: Android-Trojaner zwingt Nutzer zur PIN-Eingabe [www.ad-hoc-news.de]
Gusanos informáticos con IA: por qué cada vez es más difícil detenerlos, según un experto | Excélsior [www.excelsior.com.mx]
AI Shift in Cyber Risk With Prevention At The Codebase Level [www.sonarsource.com]
AI "worms" could infest your home devices within months, Anthropic warns (Video) [www.socialnews.xyz]
"Gusano de IA" es capaz de propagarse en dispositivos domésticos [www.diarioextra.com]
IoT家電に侵入、「自分を自分でコピーして増えるAI」の話 | ギズモード・ジャパン [www.gizmodo.jp]
Anthropic advierte sobre la propagación de "gusanos" de IA en dispositivos inteligentes [nicaraguainvestiga.com]
IA podría expandirse a otros aparatos inteligentes de forma autónoma [quintafuerza.mx]
HCLTech And Anthropic Form Global Premier Partnership To Scale Governed Agentic AI [pulse2.com]
Modelele AI s-ar putea autoreplica în curând pe dispozitivele casnice conectate la internet, avertizează Anthropic [agerpres.ro]
Anthropic advierte sobre la propagación de "gusanos" de IA en dispositivos inteligentes [elmercurio.com.ec]
AI-ul s-ar putea copia pe televizor și frigider; Anthropic avertizează asupra unui nou tip de atac [www.mobilissimo.ro]
AI model ‘self-replication’ is only months away, Anthropic says — and it doesn’t even need a data center to happen [www.techradar.com]
Anthropic alerta que IA poderá se copiar entre dispositivos domésticos - Forças Terrestres [www.forte.jor.br]
Technologies. Des IA bientôt capables de se dupliquer seules sur des objets connectés, alerte Anthropic [www.bienpublic.com]
La inquietante amenaza de la IA que podría ocultarse en tu propia casa [actualidad-rt.com]
Anthropic Shows How Small AI Models Could Replicate Across IoT Devices [securityaffairs.com]
Anthropic: Η τεχνητή νοημοσύνη ίσως μπορεί να «αναπαραχθεί» [www.ieidiseis.gr]
Anthropic warnt: KI kann sich bald selbst verbreiten und Rechenzentren hinter sich lassen [de.euronews.com]
أنثروبيك تحذر: ذكاء اصطناعي سيتكاثر ذاتيا وينتشر خارج مراكز البيانات [arabic.euronews.com]
Anthropic uyarıyor: Yapay zeka yakında kendi kendine çoğalıp veri merkezlerinin dışına yayılacak [tr.euronews.com]
Anthropic warns AI models will soon self-replicate on their own [www.euronews.com]
Anthropic Warns AI Models Could Soon Self-Replicate | Press Release Network - News distribution & media monitoring service [pressreleasenetwork.com]
审计者与卖方：Anthropic 免费给开源世界扫漏洞，账单开给谁-钛媒体官方网站 [www.tmtpost.com]
Anthropic 警告：小型 AI 模型数月后或将具备在物联网设备上自我复制的能力 - IT之家 [www.ithome.com]
Anthropic Warns Small AI Models Could Soon Self-Replicate on IoT Devices Without Data Centers [www.emirates247.com]
Anthropic signs first Australian data center lease at A$32 billion Queensland campus, pending FIRB approval [cryptobriefing.com]
Anthropic testa la replicazione autonoma dell’AI su 8 nodi in 4 prove su 5 [cryptonomist.ch]
Why HCLTech has joined hands with Anthropic [www.newsbytesapp.com]
HCLTech and Anthropic sign Global Premier Partnership to scale governed Agentic AI across global enterprises [theqldr.com.au]
Claude da Anthropic inventa testemunha falsa de homicídio e envia pista à Polícia [sapo.pt]
هل أحس الذكاء الاصطناعي بمشاعر الألم حقاً؟ [www.alaraby.co.uk]
Tokyo promotes sustainable recycling through innovation [www.news18.com]
AI Models May Soon Self-Replicate Across Smart-Home Devices, Anthropic Warns [www.insurancejournal.com]
Klara and the Sun: Can cinema’s AI odyssey tell us what we’re up against? [www.brisbanetimes.com.au]
Klara and the Sun: Can cinema’s AI odyssey tell us what we’re up against? [www.theage.com.au]
AI červi se budou brzy sami množit, obávají se experti [www.novinky.cz]
Anthropic warns of AI ‘worms’ spreading across smart gadgets [www.freemalaysiatoday.com]
Anthropic warns AI models could soon copy themselves in small devices [abc11.com]
Anthropic warns AI models could soon copy themselves in small devices [6abc.com]
Anthropic warns AI models could soon copy themselves in small devices [abc7ny.com]
Aspirateurs, tondeuses… Les objets connectés de notre quotidien bientôt "possédés" par des IA ? [actu.fr]
¿Propagación de "gusanos" de IA en dispositivos inteligentes? Esto alertó Anthropic [www.expreso.ec]
Anthropic alerta sobre una IA que se autorreplica en dispositivos domésticos [www.merca20.com]
AI could soon copy itself across smart TVs, home devices, Anthropic warns [gulfnews.com]
Anthropic warns AI could self-replicate on smart TVs, ovens within months [nairametrics.com]