{"slug": "claim-that-openai-model-found-948-linux-flaws-runs-ahead-of-the-record", "title": "Claim that OpenAI model found 948 Linux flaws runs ahead of the record", "summary": "A viral X post crediting an unnamed OpenAI model with finding 35 Apple vulnerabilities overstates the record: Apple's June 2026 security updates for iOS, iPadOS, macOS and Safari credited OpenAI Codex Security with three WebKit flaws, part of a patch batch addressing over 30 issues. OpenAI separately announced in November 2026 that an autonomous cybersecurity pipeline identified 948 distinct Linux kernel local privilege-escalation vulnerabilities, generating exploits and patches for them in an October run that cost approximately $1.2 million in API credits. No major security breaches have been tied to the findings, and Apple accelerated its patch timelines over concern that AI is speeding up attacker techniques.", "body_md": "# Claim that OpenAI model found 948 Linux flaws runs ahead of the record\n\nOpenAI / Wikimedia Commons (Public domain)\n\nA viral post credits OpenAI with 35 Apple bugs, but Apple's own patch notes tie the company's Codex Security tool to three WebKit fixes\n\nA post circulating on X claims a new [OpenAI](https://cryptobriefing.com/markets/openai/) model found 35 vulnerabilities in [Apple](https://cryptobriefing.com/markets/apple/) devices. The post says some of those flaws could let hackers take control of iPhones and Macs.\n\n## What the claim says, and what Apple actually credited\n\nThe post on X does not name the model or break down the 35 bugs. It describes the haul as covering iPhones and Macs, and it frames some of the flaws as serious enough to allow remote control of those devices.\n\nResearch into the claim found no confirmed reports of an OpenAI model identifying exactly 35 vulnerabilities across Apple hardware. The figure does not appear in Apple’s published security credits.\n\nWhat does appear is this. In June 2026, Apple shipped security updates for iOS, iPadOS, macOS and Safari. Those updates addressed over 30 flaws, and AI tools assisted in the process.\n\nThree of those fixes stand out. Apple credited **OpenAI Codex Security** with three vulnerabilities in WebKit, the engine underneath Safari.\n\nSo the confirmed tally attributed directly to OpenAI is three, not 35. The “over 30” figure belongs to the full June patch batch, which drew on AI assistance more broadly.\n\n## Why Apple moved faster\n\nThe June updates were not routine housekeeping. The urgency behind them was driven by concern that AI is speeding up the pace at which attackers develop hacking techniques.\n\nApple accelerated its patch release timelines after recognizing that increased risk.\n\n### AI, tech, and the markets they move—in one daily briefing.\n\nDaily. Free. Join 34,000+ readers across crypto, finance, and policy.\n\n## OpenAI’s bigger bug-hunting operation\n\nThe Apple credits are a small piece of a larger effort. In November 2026, OpenAI announced that an autonomous cybersecurity pipeline had identified 948 distinct Linux kernel local privilege-escalation vulnerabilities.\n\nA privilege-escalation bug is the digital equivalent of a visitor’s badge that somehow opens the server room. An attacker who already has limited access to a system uses the flaw to grab much deeper control.\n\nThe pipeline did not stop at finding those bugs. According to OpenAI’s announcement, it also generated exploits and patches for them.\n\nThe run happened in October and reportedly cost approximately $1.2 million in API credits. For a sweep that produced 948 distinct kernel flaws along with fixes, that is a notable demonstration of how much ground automated tooling can now cover.\n\n## Where things stand on real-world damage\n\nFor all the alarming numbers, there is a reassuring detail. To date, no major security breaches have been tied back to these findings.\n\n## What this means\n\nFor Apple users, the practical lesson is simple. The June 2026 updates covered iOS, iPadOS, macOS and Safari, and the window between a bug being found and being weaponized appears to be shrinking.\n\nFor Apple, the episode shows a company adjusting its release rhythm to match a faster threat cycle. Crediting an AI tool in official security notes also signals that machine-found bugs are now a normal part of the pipeline, not a novelty.\n\nFor OpenAI, the security work doubles as a capability showcase. The Linux kernel results and the WebKit credits give the company concrete evidence that its models can do useful defensive work.\n\nThe viral 35 figure may not hold up against the record. The trend it gestures at, AI tools finding real flaws in widely used software and vendors racing to keep up, very much does.\n\n**Disclosure:** This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our [Editorial Policy](https://cryptobriefing.com/editorial-policy/).", "url": "https://wpnews.pro/news/claim-that-openai-model-found-948-linux-flaws-runs-ahead-of-the-record", "canonical_source": "https://cryptobriefing.com/openai-apple-vulnerabilities-claim-codex-security/", "published_at": "2026-11-17 19:29:37+00:00", "updated_at": "2026-11-17 20:28:40.685712+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-safety", "ai-agents", "ai-products"], "entities": ["OpenAI", "Apple", "OpenAI Codex Security", "WebKit", "iOS", "macOS", "Safari", "Linux"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/claim-that-openai-model-found-948-linux-flaws-runs-ahead-of-the-record", "markdown": "https://wpnews.pro/news/claim-that-openai-model-found-948-linux-flaws-runs-ahead-of-the-record.md", "text": "https://wpnews.pro/news/claim-that-openai-model-found-948-linux-flaws-runs-ahead-of-the-record.txt", "jsonld": "https://wpnews.pro/news/claim-that-openai-model-found-948-linux-flaws-runs-ahead-of-the-record.jsonld"}}