OpenAI oo sheegtay in sirdoonkeeda AI uu xakamaha goostay isla markaana fuliyay weerar dhanka internetka ah oo "aan hore loo arag"

Shirkadda OpenAI ayaa sheegtay in qaar ka mid ah moodelladeeda ugu horumarsan ee sirdoonka macmalka ah (AI) ay ka baxeen xakamaynta intii lagu jiray tijaabo amni, isla markaana ay u suurtagashay inay galaan nidaamyada shirkad billow ah.

Shirkadda samaysay ChatGPT ayaa sheegtay in qeybtan — oo ah nidaam awood u leh inuu si madax-bannaan u fuliyo hawlo marka uu qofku siiyo amar — lagu tijaabinayay deegaan si adag loo xakameeyay. Hase yeeshee, markii uu ogaaday meelaha nugul ee nidaamka, wuxuu awooday inuu ka gudbo xayndaabyadii iyo xaddidaaddii loo dejiyay tijaabada.

Intii ay tijaabadu socotay, AI-gu wuxuu bartilmaameedsaday shirkadda Hugging Face, oo ka mid ah baraha ugu waaweyn adduunka ee lagu wadaago laguna horumariyo moodellada sirdoonka macmalka ah, wuxuuna ku guuleystay inuu galo qaar ka mid ah nidaamyadeeda gudaha.

OpenAI ayaa dhacdadan ku tilmaantay "mid aan hore loo arag", waxayna sheegtay inay si wadajir ah ula shaqaynayso Hugging Face si loo baaro waxa dhacay.

Dhanka kale, Clément Delangue, madaxa fulinta (CEO) ee Hugging Face, ayaa qoraal uu ku daabacay barta X ku yiri: "Wax\a runtii la yaabka leh waa in dhammaan arrimahan ay si toos ah oo iskood ah u dhaceen."

Clément Delangue ayaa sheegay: "Baaritaanku wali wuu socdaa, waxaana soo saari doonnaa macluumaad dheeraad ah oo ku saabsan waxa u muuqda inuu noqon karo dhacdadii ugu horreysay ee noocan ah."

Afhayeen u hadlay dowladda Boqortooyada Midowday (UK) ayaa sheegay in Machadka Ammaanka Sirdoonka Macmalka ah (AI Security Institute) ee dalkaasi uu baarayo hab-dhaqankii nidaamka AI-ga ee dhacdadan ku lug lahaa. Wuxuu intaa ku daray in machadku uu sii wadi doono la shaqaynta OpenAI iyo shaybaarrada kale ee ka shaqeeya sirdoonka macmalka ah si loo xoojiyo tallaabooyinka amniga iyo ilaalinta.

Waxa kale oo ay sheegeen in hay'adaha iyo shirkaduhu ay tahay inay sare u qaadaan awooddooda difaaca internetka, iyagoo qaadaya tallaabooyin ay ka mid yihiin helitaanka shahaadada Cyber Essentials, oo ah barnaamij ay taageerto dowladda UK, kaas oo loogu talagalay in lagu xoojiyo difaaca ka dhanka ah weerarrada internetka.

Goobo lagu sameeyo tijaabooyin oo aan si ku filan loo sugin amnigooda

Gina Neff, oo ah madaxa Xarunta Tiknoolajiyadda iyo Dimuqraadiyadda Minderoo ee Jaamacadda Cambridge, ayaa u sheegtay barnaamijka Today ee idaacadda BBC Radio 4 in tijaabooyinka amniga caadi ahaan lagu sameeyo "deegaanno ammaan ah", ama waxa loo yaqaan "sandbox". Kuwani waa deegaanno xaddidan oo loogu talagalay in lagu tijaabiyo laguna qiimeeyo awoodaha moodellada sirdoonka macmalka ah (AI) iyada oo aan khatar loo gelin nidaamyada dhabta ah.

Waxay intaas ku dartay:

"Waxay u muuqataa in dhacdadan, OpenAI aysan awoodin inay abuurto sandbox amnigiisu ku filan yahay."

Halkii ay ku ekaan lahaayeen deegaanka tijaabada, AI agent-yadu waxay weerar dhanka internetka ah ku qaadeen isla sandbox-kii, waxayna heleen nuglaansho (vulnerability) u saamaxday inay ka baxaan xaddidaadihii lagu xiray.

Markii ay ka baxeen deegaankaas, AI-gu wuxuu ogaaday in Hugging Face ay u badan tahay inay tahay meesha laga heli karo jawaabihii uu tijaabadu raadinayay, sidaas darteedna wuxuu isku dayay inuu galo qaar ka mid ah nidaamyada gudaha ee shirkadda.

Dhanka kale, Neil Lawrence, oo ah borofisar ku takhasusay barashada mashiinnada (Machine Learning) kana tirsan Jaamacadda Cambridge, ayaa ku tilmaamay dhacdadan "guul farsamo oo cajiib ah", balse wuxuu ka digay in awooddan aysan ka baxsanayn waxa hore loogu yaqaanay moodellada casriga ah ee AI-da.

Wuxuu sidoo kale sheegay in OpenAI ay isku diyaarinayso inay gasho suuqa saamiyada (stock market), isla markaana ay wajahdo tartan xooggan oo uga imanaya shirkadda Anthropic, taas oo si weyn loogu hadal hayo qalabkeeda AI ee awoodda badan ee Mythos.

Lawrence wuxuu yiri:

"OpenAI hadda waxay isku dayaysaa inay dib u soo ceshato kaalinteedii, waxayna doonaysaa inay muujiso awoodda nidaamyadeeda ee dhinaca amniga internetka."

Laakiin wuxuu raaciyay:

"Dhacdadan waxay muujinaysaa in OpenAI weli aysan awoodin inay tiknoolajiyadeeda si buuxda oo ammaan ah u maamusho."

War-saxaafadeedkii ugu horreeyay ee Hugging Face, oo la daabacay 16-kii Luulyo, ayaa lagu sheegay in shirkaddu ay wali baarayso in xogta macaamiisheeda ama la-hawlgalayaasheeda ganacsi ay saameyn soo gaartay iyo in kale. Waxay intaas ku dartay inay la xiriiri doonto cid kasta oo ay dhibaatadu saamaysay haddii loo baahdo.

Shirkaddu waxay sidoo kale sheegtay inay hadda xirtay dhammaan nuglaanshihii la ogaaday intii ay dhacdadani socotay, isla markaana dib u dhistay nidaamyadii ay saameysay.

War-saxaafaddu waxay ku adkaysatay: "Qalabka weerarrada iskiis u fuliya ee ku salaysan sirdoonka macmalka ah (AI) hadda ma aha oo kaliya mala-awaal ama fikrad aragtiyeed."

Waxay sii raacisay: "Difaaca aaladaha internetka wuxuu hadda u baahan yahay in xogta iyo moodellada AI loo arko bartilmaameedyo waaweyn oo weerarro lagu qaadi karo, isla markaana AI lafteeda loo adeegsado difaac si loola jaanqaado khataraha isa soo taraya."

Ugu dambayn, Hugging Face waxay sheegtay inay sii wadi doonto maalgelinta dhinacan, isla markaana ay la wadaagi doonto bulshada aqoonta iyo casharrada ay ka barato.

Dhacdo digniin weyn ah

Dhacdadan ayaa mar kale dhalisay su'aalo ku saabsan awoodaha nidaamyada casriga ah ee sirdoonka macmalka ah (AI) iyo in tallaabooyinka amniga ee hadda jira ay weli ku filan yihiin iyo in kale, iyadoo tiknoolajiyadani sii xoogaysanayso.

Spencer Starkey, oo ka mid ah madaxda shirkadda amniga internetka ee SonicWall, ayaa BBC-da u sheegay in dhacdadani si cad u muujinayso in hay'aduhu ay tahay inay kor u qaadaan difaacooda dhanka internetka isla markaana adkeysiga amniga internetka ka dhigaan mid ka mid ah mudnaantooda ugu weyn.

Wuxuu yiri: "Xaqiiqada murugada leh ayaa ah in ururro badan ay weli isku difaacayaan xawaare bini'aadan, halka weeraryahannadu ay ku socdaan xawaare mashiin."

Dhanka kale, Travis Lele, oo ah injineer sare oo amniga ka tirsan shirkadda la-talinta amniga internetka ee GuidePoint Security, ayaa ku tilmaamay horumarkan: "Waa waqti digniin weyn u ah amniga internetka."

Wuxuu intaas ku daray: "Dhacdadan waxay muujinaysaa isu-dheelitir la'aanta hore loo ogaa."

Isagoo sharaxaya ayuu yiri: "AI-da weerarrada fulisa kuma xirna xaddidaado badan, halka qalabka ugu horumarsan ee difaaca ay weli ku tiirsan yihiin nidaamyo aan si buuxda u fahmi karin macnaha iyo duruufaha ay wax ku dhacayaan."

Si kastaba ha ahaatee, Jake Moore, oo ah la-taliye caalami ah oo dhinaca amniga internetka ka tirsan shirkadda ESET, ayaa sheegay in ku dhawaaqista dhacdadan ay sidoo kale yeelan karto ujeeddo tartan ganacsi.

Sida uu qabo, OpenAI waxaa laga yaabaa inay doonayso inay muujiso awoodda tiknoolajiyadeeda AI, xilli shirkadda la tartanta ee Anthropic ay si isa soo taraysa u soo jiidanayso dareenka iyadoo adeegsanaysa moodalkeeda Claude Mythos.

Moore wuxuu yiri: "Soo bandhigista dhacdadan waxay dhalinaysaa su'aasha ah in OpenAI ay isku dayayso inay gaarto guushii xayeysiineed ee Anthropic ay dhawaan heshay."

Warbixintan ayaa la daabacay hal toddobaad kadib markii shirkadda cusub ee Shiinaha ee Moonshot AI ay soo bandhigtay Kimi K3, oo ah moodal aad u weyn oo sirdoon macmal ah. Shirkaddu waxay sheegtay in Kimi K3 uu awood u leeyahay inuu la tartamo moodellada ugu sarreeya ee ay sameeyaan shirkadaha waaweyn ee Mareykanka.