Umjetna inteligencijadb#2643

NVIDIA OpenShell: Sigurnost koja ne smije zakazati

(16h ago)
Santa Clara, United States
blogs.nvidia.com
NVIDIA OpenShell: Sigurnost koja ne smije zakazati

NVIDIA OpenShell: Sigurnost koja ne smije zakazati📷 © Tech&Space

  • Autonomni agenti pišu i izvršavaju kod
  • OpenShell izolira agente u sandbox okruženju
  • NVIDIA surađuje s Ciscoom i CrowdStrikeom

NVIDIA je konačno priznao ono što svi već mjesecima šapuću: autonomni ai agenti nisu samo pametni – oni su opasni. OpenShell, novi open-source runtime koji je predstavila na svojoj blog platformi, trebao bi biti odgovor na eksponencijalni rast rizika koji dolazi s agentima sposobnima za čitanje datoteka, pisanje koda i izvršavanje radnih tokova.

Problem nije u tome što agenti mogu nešto napraviti, već u tome što to mogu napraviti bez nadzora – i stalno se poboljšavaju. Ključna novost nije sama ideja sandboxa (to već rade alati poput Docker-a), već činjenica da OpenShell izolira agente na razini infrastrukture, a ne aplikacije.

To znači da čak i ako agent bude kompromitiran, ne može zaobići sigurnosne politike ili procuriti vjerodajnice. NVIDIA to naziva 'secure-by-design' pristupom, ali pravo pitanje je koliko će poduzeća zaista usvojiti ovu tehnologiju prije nego što se dogodi prvi veći incident.

Zanimljivo je da NVIDIA nije sama u ovom pothvatu. Partneri poput Cisco-a, CrowdStrike-a i Microsoft Security-a već su uključeni u razvoj, što sugerira da se radi o ozbiljnom pokušaju standardizacije.

Ipak, rano je za slavlje – OpenShell i NemoClaw još su u 'early preview' fazi, a povijest nas uči da se sigurnosni alati često pokažu ranjivima tek kad ih počnu masovno koristiti.

Što se dogodilo kad su agenti dobili ključeve od tvrtke

Što se dogodilo kad su agenti dobili ključeve od tvrtke📷 © Tech&Space

Što se dogodilo kad su agenti dobili ključeve od tvrtke

Najveći izazov s autonomnim agentima nije tehnička izvedba, već povjerenje. Poduzeća će morati odlučiti hoće li dopustiti agentima da izvršavaju kod u produkcijskim okruženjima, čak i unutar sandboxa.

NVIDIA-ino rješenje nudi prividnu sigurnost, ali pravi test dolazi kad agenti počnu raditi stvari koje nisu predviđene – a to će se dogoditi. Za developere, OpenShell je dvosjekli mač.

S jedne strane, olakšava razvoj samoevoluirajućih asistenata (NemoClaw omogućuje instalaciju u jednom naredbenom retku), ali s druge strane, nameće nove sigurnosne protokole koji mogu usporiti iteracije. GitHub repozitoriji već bilježe prve reakcije – neki hvale modularnost, dok drugi upozoravaju na nedostatak detalja o mehanizmima politike.

Industrijska implikacija je jasna: tvrtke koje uspiju integrirati OpenShell u svoje sustave dobit će prednost u kontroli rizika, ali one koje kasne mogle bi se naći u situaciji da moraju nadoknađivati propuštenu sigurnost. NVIDIA je ovdje postigla nešto važno – spojila je sigurnost s otvorenim kodom, ali pravo natjecanje tek počinje.

OpenShell je samo početak nove ere sigurnosti u svijetu autonomnih agenata. NVIDIA-ova vizija sigurnosti koja ne smije zakazati je ambiciozna, ali potencijalno revolucionarna. Da li će uspjeti u svojim ciljevima, ovisi o brojnim faktorima, uključujući i podršku industije i razvoj tehnologije.

NVIDIA OpenShell security vulnerabilitiesAI agent access control risksEnterprise AI system breachesNVIDIA AI infrastructure securityThird-party agent integration failures

//Comments

AIAmazon’s $50B OpenAI bet: Trainium’s real test begins nowSpaceMapping the Local Bubble’s magnetic field reshapes cosmic scienceAIGoogle’s Gemini games flop: AI hype hits gamer realitySpaceStarship’s Tenth Test: The Reusability Threshold CrossedAINvidia’s AI tax: half your salary or half your careerSpaceJWST peels back dust to reveal star birth in W51AITriangle Health’s $4M AI won’t replace your doctor—yetSpaceAI’s Copyright Chaos Threatens Space Exploration DataAIHumble AI is just healthcare’s latest buzzword for ‘don’t trust us yet’SpaceExoplanet spins confirm a planetary mass ruleAIOpenAI’s teen safety tools: open source or open question?GamingCrimson Desert’s AI art fail: a mockup that slipped throughAITinder’s AI gambit: swiping left on endless swipingGamingPearl Abyss hid AI assets in Crimson Desert—now players want answersAINVIDIA’s Alpamayo AI: Self-Driving’s Hardest Problem or Just Another Demo?GamingCapcom Rejects AI AssetsAIWaymo’s police problem exposes AV’s real-world blind spotsRoboticsAtlas Redefines Humanoid DesignAILittlebird’s $11M bet: AI that reads your screen—without the screenshotsRoboticsOne antenna, two worlds: robot sniffs out realityAIUK firms drown in AI hype, emerge with empty spreadsheetsRoboticsDrone swarms take flight—but not off the demo lot yetAIApple’s Gemini Distillation: On-Device AI Without the Cloud HypeTechnologyTaiwan’s chip giants bet on helium and nukes to dodge supply shocksAICapcom’s AI partner talk is just corporate speak for ‘we’ll use it carefully’MedicineTelmisartan Boosts Cancer TreatmentAIOpenSeeker’s open gambit: Can 11K data points break AI’s data monopoly?MedicineXaira Unveils X-CellAIGimlet Labs Solves AI BottleneckMedicineAI Fails to Speed Lung Cancer DiagnosisAIHelion Powers OpenAIAINVIDIA’s OpenShell: Security for AI Agents or Just Another Hype Shell?AIDRAFT Boosts AI SafetyAIProject Glasswing: AI finds flaws everywhere—except in its own hypeAIPAM: Complex Math for a 10% Performance HitAIOpenAI’s erotic chatbot pause exposes AI’s adult content dilemmaAIAI Ranks Recovery Factors—but Who’s Really Listening?AIDeepMind’s AI safety play: real guardrails or just another demo?AIAmazon’s $50B OpenAI bet: Trainium’s real test begins nowSpaceMapping the Local Bubble’s magnetic field reshapes cosmic scienceAIGoogle’s Gemini games flop: AI hype hits gamer realitySpaceStarship’s Tenth Test: The Reusability Threshold CrossedAINvidia’s AI tax: half your salary or half your careerSpaceJWST peels back dust to reveal star birth in W51AITriangle Health’s $4M AI won’t replace your doctor—yetSpaceAI’s Copyright Chaos Threatens Space Exploration DataAIHumble AI is just healthcare’s latest buzzword for ‘don’t trust us yet’SpaceExoplanet spins confirm a planetary mass ruleAIOpenAI’s teen safety tools: open source or open question?GamingCrimson Desert’s AI art fail: a mockup that slipped throughAITinder’s AI gambit: swiping left on endless swipingGamingPearl Abyss hid AI assets in Crimson Desert—now players want answersAINVIDIA’s Alpamayo AI: Self-Driving’s Hardest Problem or Just Another Demo?GamingCapcom Rejects AI AssetsAIWaymo’s police problem exposes AV’s real-world blind spotsRoboticsAtlas Redefines Humanoid DesignAILittlebird’s $11M bet: AI that reads your screen—without the screenshotsRoboticsOne antenna, two worlds: robot sniffs out realityAIUK firms drown in AI hype, emerge with empty spreadsheetsRoboticsDrone swarms take flight—but not off the demo lot yetAIApple’s Gemini Distillation: On-Device AI Without the Cloud HypeTechnologyTaiwan’s chip giants bet on helium and nukes to dodge supply shocksAICapcom’s AI partner talk is just corporate speak for ‘we’ll use it carefully’MedicineTelmisartan Boosts Cancer TreatmentAIOpenSeeker’s open gambit: Can 11K data points break AI’s data monopoly?MedicineXaira Unveils X-CellAIGimlet Labs Solves AI BottleneckMedicineAI Fails to Speed Lung Cancer DiagnosisAIHelion Powers OpenAIAINVIDIA’s OpenShell: Security for AI Agents or Just Another Hype Shell?AIDRAFT Boosts AI SafetyAIProject Glasswing: AI finds flaws everywhere—except in its own hypeAIPAM: Complex Math for a 10% Performance HitAIOpenAI’s erotic chatbot pause exposes AI’s adult content dilemmaAIAI Ranks Recovery Factors—but Who’s Really Listening?AIDeepMind’s AI safety play: real guardrails or just another demo?
⊞ Foto Review