Google has reported that it fixed more security vulnerabilities in the Chrome browser during June than in the preceding two years combined, attributing the surge in bug fixes to the integration of artificial intelligence (AI) tools within its development process. According to a report by TechCrunch, the company patched 1,072 security flaws in the last two Chrome versions, released in June, surpassing the total of 1,036 fixes from the prior 23 versions over the past two years. This marks a dramatic acceleration in the rate of vulnerability resolution, driven by the deployment of large language models (LLMs) such as Gemini. The shift toward AI-driven security practices appears to be reshaping the landscape of cybersecurity. Google's internal data, presented in a white paper, highlights an exponential growth in the number of bugs detected and resolved, with the latest two Chrome milestones, versions 149 and 150, showcasing a sharp uptick compared to earlier releases. The company attributes this progress to the use of AI models that enable proactive identification and mitigation of vulnerabilities. Doug Turner, director of engineering at Chrome, emphasized that these models have fundamentally altered the economics of cybersecurity, enabling the automation of processes that once required extensive manual effort. This trend is not unique to Google. Microsoft recently announced that it addressed a record 570 security flaws across its product suite during its recent Patch Tuesday update, citing AI as a contributing factor. The company introduced MAI-Cyber-1-Flash, a cybersecurity model integrated with GPT-5.4 within its MDASH framework, which reportedly outperformed competing models from Anthropic, Google, and OpenAI on standard benchmarks. Microsoft's new security platform, named Perception, employs agentic systems to simulate attack scenarios, detect vulnerabilities, and apply corrective measures, aiming to streamline the entire security workflow. Meanwhile, Anthropic's AI model, Mythos, has been instrumental in uncovering a vast array of security flaws across Microsoft's software ecosystem. During a meeting held in mid-May, Microsoft engineers discussed the challenges posed by Mythos, which identified hundreds of critical and important bugs in SharePoint and other applications. Despite the rapid discovery of vulnerabilities, Microsoft has struggled to keep pace with the volume of findings, leading to a scramble among developers to address the most severe issues first. Internal documents suggest that the company intends to eventually tackle moderate-severity bugs, though low-severity flaws remain largely unaddressed. The implications of this AI-driven vulnerability discovery extend beyond individual companies. A recent report by ProPublica highlighted concerns that the proliferation of AI tools among adversaries could soon erode the competitive advantage currently enjoyed by Western tech firms. National security experts warn that the window of opportunity for proactive security improvements is rapidly narrowing, with the Five Eyes intelligence alliance issuing warnings about the imminent threat posed by state-sponsored AI-based cyber operations. In response to these developments, Apple has yet to demonstrate a comparable escalation in the number of security fixes. Independent analysis suggests that the company has patched approximately 482 bugs in 2026, placing it on track to match or exceed the figures from the previous year. However, this rate of improvement falls short of the exponential growth observed in Google and Microsoft's approaches. TechCrunch attempted to reach out to Apple for further clarification but received no response. As the race to secure digital infrastructure intensifies, the role of AI in both offensive and defensive cyber strategies continues to evolve. With each major tech firm investing heavily in AI-powered security solutions, the balance between innovation and risk management becomes ever more precarious. The coming months will likely see further refinements in how these technologies are deployed, with the ultimate aim of staying ahead of emerging threats in an increasingly interconnected and vulnerable digital world.
5 poročil
TechCrunchNeodvisenSredinaDejstva 85Objektivnost 80predvčerajšnjimGoogle trdi, da je njegov brskalnik Chrome samo v juniju 2024 popravil več varnostnih napak kot v prejšnjih dveh letih skupaj, s čimer je povečal uporabo orodij AI, kot je Gemini. Podjetje je poročalo, da je v zadnjih dveh posodobitvah Chromea popravilo 1072 varnostnih napak, kar je preseglo skupno 1036 popravil iz prejšnjih 23 različic. To pomeni pomemben premik v praksi kibernetske varnosti, saj AI omogoča hitrejše in obsežnejše odkrivanje in reševanje ranljivosti.
Ocena pristranskosti (Sredina): V članku je predstavljen dejanski poročilo o tehnoloških napredkih na področju kibernetske varnosti brez očitnega ideološkega nagibanja.
Zakaj dejstva (85): The article accurately reflects the information from the primary source document, mentioning the use of AI to find and fix more bugs in Chrome than in the past two years combined. It references specific numbers (1,072 bugs fixed in June) and aligns with the timeline and goals outlined in the primary
Zakaj objektivnost (80): The tone remains neutral and informative, focusing on the impact of AI on Chrome security. There is no overt bias or emotional language, though it emphasizes the significance of the achievement without delving into potential risks or limitations.
QuartzNeodvisenSredinaDejstva 75Objektivnost 70pred 4 dneviMicrosoft je najavil začetek svojega prvega namenskega modela kibernetske umetne inteligence MAI-Cyber-1-Flash. Ta model je integriran z GPT-5.4 v okviru Microsoftovega MDASH in je pokazal vrhunske zmogljivosti v primerjavi s podobnimi modeli, ki so jih razvili Anthropic, Google in OpenAI na pomembnem merilnem testu. Razvoj poudarja Microsoftov naraščajoči poudarek na rešitvah kibernetske varnosti, ki jih poganja napredna tehnologija AI. Model je postavljen kot bolj stroškovno učinkovita alternativa konkurenčnim ponudbam na trgu.
Ocena pristranskosti (Sredina): Članek obravnava tehnološki napredek Microsofta na področju kibernetske varnosti AI brez kakršnih koli očitnih političnih komentarjev, okvirov ali pristranskosti.
Zakaj dejstva (75): The article discusses Microsoft's launch of a new cybersecurity model and platform, referencing performance benchmarks against competitors. While it doesn't directly reference the Chrome-specific updates in the primary source, it aligns with the broader theme of AI-driven cybersecurity improvements.
Zakaj objektivnost (70): The tone is promotional and forward-looking, highlighting Microsoft's achievements relative to competitors. While not overtly biased, it frames the release in a competitive light, suggesting a strategic advantage over other AI models.
ProPublicaNeodvisenSredinaDejstva 70Objektivnost 65pred 3 dneviProPublica poroča, da model umetne inteligence Anthropic, Mythos, prepozna programske ranljivosti v izdelkih Microsofta z brez primere hitrostjo, ki presega zmožnost Microsofta, da jih obravnava. Med srečanjem maja so inženirji Microsofta priznali, da se borijo, da bi sledili obsegu kritičnih napak, ki jih je odkril Mythos, zlasti v SharePointu.
Ocena pristranskosti (Sredina): Medtem ko članek obravnava kibernetsko varnost in zmogljivosti umetne inteligence, ki imajo geopolitične posledice, ostaja okvir uravnotežen.
Zakaj dejstva (70): The article discusses Microsoft's struggle to keep up with vulnerabilities found by Anthropic's Mythos AI, but it lacks direct alignment with the primary source document. It focuses on a different company (Microsoft) and a different AI model (Mythos), which may not be directly related to the Chrome-
Zakaj objektivnost (65): The article presents a somewhat negative tone toward Microsoft's ability to handle the volume of vulnerabilities, implying urgency and pressure. While not overtly biased, it frames the situation in a way that suggests a challenge rather than a balanced assessment.
Antropski Claude AI model najde napake v algoritmih za šifriranje, ki jih je težko razbitiModel AI, ki ga je razvil Anthropic, znan kot Claude Mythos Preview, je med testiranjem ugotovil ranljivosti v nekaterih kriptografskih algoritmih. Ti algoritmi se uporabljajo za zaščito spletnih finančnih transakcij, zasebnih komunikacij in drugih občutljivih podatkov. Odkritje poudarja potencialne slabosti v šifrirnih metodah, ki se štejejo za težko razbiti.
Ocena pristranskosti (Sredina): Članek obravnava tehnološki napredek, povezan s kriptografijo, ne da bi se odločil za politična vprašanja.
Zakaj dejstva (65): The article mentions an Anthropic AI model discovering flaws in encryption algorithms, but it does not connect this to the Chrome-specific updates or the broader AI-driven security improvements discussed in the primary source. It provides limited context and does not reference the primary source doc
Zakaj objektivnost (60): The tone is more technical and focused on the discovery of flaws, but it lacks balance by emphasizing the severity of the findings without addressing the broader context of AI-driven security improvements. The language leans slightly toward highlighting the importance of the discovery.
TechCrunchNeodvisenSredinaDejstva 40Objektivnost 85pred 5 dnevi Microsoft predstavi prvi model kibernetske varnosti in nov sistem kibernetske varnostiMicrosoft je predstavil svoj prvi model AI, ki je osredotočen na kibernetsko varnost, MAI-Cyber-1-Flash, skupaj z novo platformo za kibernetsko varnost, ki jo poganja AI, imenovano Perception. Ti orodji si prizadevajo za izboljšanje varnosti podjetij z avtomatizacijo nalog, kot so prepoznavanje in popravljanje ranljivosti programske opreme. Model MAI-Cyber-1-Flash naj bi presegel konkurente, kot so Google's Gemini, OpenAI's GPT različice in Anthropic's Mythos na uveljavljenih merilih.
Ocena pristranskosti (Sredina): Članek obravnava tehnološki napredek v kibernetski varnosti brez kakršnega koli očitnega političnega okvirja, pristranskosti ali poudarka na političnih osebnostih, politikah ali ideoloških perspektivah.
Zakaj dejstva (40): The article focuses on Microsoft launching a new cybersecurity model and platform, which is unrelated to the primary source document about Microsoft's financial results. It does not reference or align with the earnings data, making it factually disconnected from the main topic.
Zakaj objektivnost (85): The article maintains a neutral tone, presenting technical details about Microsoft's new products without apparent bias. It avoids emotionally charged language and sticks to factual descriptions of product features.
★
Ohranimo novice poštene.
ObjectiveNews financirajo bralci in je brez oglasov – pristranskost vam pokažemo, ne skrijemo. Podprite neodvisno novinarstvo za 5 €/mesec.
Postani podpornik
United States
