OpenAI says one automated scan of the Linux kernel produced 948 distinct root-level flaws for $1.2 million, and the video examined here states it correctly. In the same week SUSE tied a 10% staff cut to AI and Proton published a manifesto on AI-written code. The checkable numbers sit on the AI side, while the Linux software ports remain promises, as the video itself concedes.
Table of Contents
OpenAI says its automated scan of the Linux kernel produced 948 distinct local privilege escalation flaws, bugs that let an unprivileged user become root, at a total cost of $1,213,430 in API credits. That works out to roughly $1,279 per flaw, a price OpenAI sets against the $18.9 million to $66.3 million the same bugs would have fetched at historical Pwn2Own and TyphoonPWN payouts. The figures come from OpenAI’s own post of 4 November 2026, and the company says it has already sent hundreds of reports and patches to kernel maintainers and will send the rest in stages.
A weekly Linux news video chapter-listed the result as “about 1000 flaws”, yet its narration gives the exact figures: 1,724 avenues, 948 with working exploits, and $1.2 million in credits. That episode of Linux Weekly News also covers Affinity, Proton, Cutefish, SUSE and Brave, which makes it a good test case, because nearly every item carries a claim that can be checked against documents. The narration holds up well against the reporting, and its caveats about the software ports match what the sources show.
The kernel total is 948 and the run had four other numbers
OpenAI’s pipeline narrowed the kernel to code that low-privileged users can reach, flagged 2,308 candidate bugs and reproduced 1,724 of them. ITPro reports 1,726, a two-bug gap that the other coverage does not share. The agents then wrote 1,169 working exploits, and OpenAI says 948 counts the separate patches for the bugs behind them, so the headline is a count of distinct vulnerabilities, not of exploit programs. The exploits were tested on virtual machines with an unmodified desktop configuration and common protections switched on. The video’s narration gets this right: it separates reproduced bugs from working exploits, notes that these are local rather than network flaws and states the $1.2 million cost.
A hardened target gave far less. Pointed at an adapted version of Google’s kernelCTF, the pipeline needed 112 hours and about $33,000 to deliver 21 exploits. A late-July scan with an older model found 166 distinct bugs for $513,690, and OpenAI says October found more than five times as many new ones even after three months of kernel hardening. OpenAI’s post calls the October model “upcoming”, while ITPro reports it as GPT-6 Cyber, unveiled on 5 November.
SUSE ties a 10% cut to AI while its owner weighs a sale
On Monday 2 November 2026 SUSE chief executive Dirk-Peter van Leeuwen told staff in a memo, first reported by The Register, that the company would cut about 10% of its workforce. SUSE has more than 1,900 employees, which puts the number near 190 people, though the company has not published one. He wrote that “the requirement for some existing roles has greatly diminished”, and the memo names automation and AI as the cause. Affected staff were due to hear by 4 November, and the rest of the company gets the fiscal 2027 strategy on 16 November.
The memo is one explanation among several that fit the facts. SUSE had already asked long-serving staff in September whether they would take voluntary separation. Its owner, the Swedish private equity firm EQT, bought it from Micro Focus for $2.5 billion in 2018, listed it in Frankfurt in 2021 and took it private in 2023, and Reuters reported in March that EQT was weighing a sale. On 27 October, five days before the memo, SUSE announced agentic AI features for its Linux management tools. No coverage reviewed says which roles go, so AI as the cause is the chief executive’s account and not an established fact.
Proton’s manifesto arrived with a 90% figure and a denial
Proton chief technology officer Bart Butler wrote an internal memo titled AI @ Proton, Engineering Manifesto, which 404 Media published on 29 October 2026 and which Proton itself posted on 1 November. Its core instruction is that an engineer’s job is now to help the model write better code faster. Developer Flavio Copes, summarising 404 Media’s reporting of meeting audio, says Butler put AI-written code at about 90% of Proton’s October output and that employees said much of it came from Anthropic’s Claude. That claim passes through two layers of reporting, and the coverage reviewed does not show Proton confirming it.
The same reporting says chief executive Andy Yen dismissed a request for a few days of penetration testing as “loser thinking” and that staff would be re-interviewed about AI use with jobs at stake. Proton’s reply, covered by TechRadar on 2 November, promised no re-interviews, no AI-related job losses and no software on laptops counting AI-written code, plus publication of its code and security review practice. 404 Media had reported the tracking software, so the two accounts contradict each other, and neither side’s evidence is public. TechRadar also notes that the apology never says what Proton regrets: its treatment of staff, the security risk, or both.
Affinity and Cutefish are promises without dates
Affinity’s Linux version rests on a reply from the company’s official X account on 3 November: development is active, with no release date and no stated tooling approach, according to OMG! Ubuntu. Affinity is the free design suite owned by Canva, which bought its British developer Serif in 2024, and its iPad edition launched the same day. Linux users run it through Wine with an unofficial wrapper, and OMG! Ubuntu found problems with Canva sign-in, hardware acceleration, UI scaling and updates. OpenNET adds that Affinity earns money through paid AI features and Canva services, so the Linux build arrives inside a commercial plan.
Cutefish is further from finished. Its compatibility layer Reef reimplements macOS frameworks, and Linuxiac reports it runs NetNewsWire, Ghostty, CotEditor, Keka and Ookla Speedtest. The project says Reef holds about 460,000 lines of its own code across more than 80 frameworks, yet Reef is not open source, the Cutefish 1.0 images are nightly development builds, and no date for users exists. Five demonstrated apps make a first milestone, not a general fact that macOS apps run on Linux.
Brave’s lawsuit names five AI customers that are not defendants
The New York Times sued Brave Software in Manhattan federal court on 2 November 2026, and the publishers of the New York Daily News, the Chicago Tribune and the Orlando Sentinel filed a similar case hours later, The Next Web reports. Both centre on the Brave Index, which the publishers say holds more than 30 billion pages and is sold through a Search API to Anthropic, Perplexity, Mistral, xAI and Cohere. None of those AI companies is a defendant, and the publishers’ statements are allegations that no court has decided.
Brave is not a newcomer to this fight: it sued News Corp in March 2025 to have its indexing declared fair use, and News Corp countersued in July. The video’s chapter title states the lawsuit’s theory, selling publishers’ content to AI providers, in the lawsuit’s own terms. That is accurate as a description of the claim and silent on whether the claim will succeed.
Sponsors, gaps and the items left unchecked
The video is sponsored by Internxt, with a second sponsor segment from Tuxedo Computers, and its host says he has a monthly commercial relationship with Proton, which he states before giving his view of the manifesto. He tested Cutefish’s Reef himself and says almost nothing runs, which matches Linuxiac’s description of an early milestone. His opinions, that AI-led restructuring is short-sighted and that Brave is untrustworthy, are his own and not findings. The chapters on Linux Mint, Nvidia’s Wayland driver update, Discord’s move from Electron to Rust, Hadron and Fedora 45 were not tested against outside sources.
Three open questions matter most. Whether SUSE’s cuts are caused by AI or only explained by it. Whether Proton’s 90% figure and its tracking denial survive publication of its review process. And whether OpenAI’s kernel result holds on actively hardened kernels, where the 21 kernelCTF exploits suggest the yield is far smaller than 948 against current mainline code.
Next steps for kernel admins, SUSE customers and designers
OpenAI’s own warning is the practical part: a low-privileged account on a system missing the latest patches should be treated as equal to root against a well-resourced attacker. Linus Torvalds wrote on 8 November that “fixes are now coming in faster than the new reports”, per ITPro, which makes patch cadence on shared machines the control to check first. SUSE customers have a dated checkpoint in the 16 November strategy presentation, since the memo names no roles. Proton users can judge the apology by whether the promised code and security-review publication appears. Designers who need Affinity should keep planning around Wine, because a confirmed port with no date is not a schedule.
The video is right that Linux is gaining commercial and compatibility software. The evidence this week sits on the AI side, with numbers, dates and a bill, while the ports remain announcements.
Author:
Jan Bielik
CEO & Founder of Webiano Digital & Marketing Agency
This article is an original analysis supported by the sources cited below
Affinity comes to Linux, Proton explains recent issues, macOS apps run on Linux | Linux Weekly News
The video under examination, whose narration states the 948 and 1,724 figures, the sponsor segments and the host’s disclosed Proton relationship.
Affinity comes to Linux, Proton explains recent issues, macOS apps run on Linux | Linux Weekly News – Lemmy
Reproduces the video’s description with its eleven news chapters, two sponsor segments and the source links it cites.
AI Agents Got Root on Linux via 948 Kernel Flaws in OpenAI Research
Established the pipeline counts, the $1,213,430 cost, the kernelCTF result and OpenAI’s bounty comparison and patching figures.
OpenAI says its new cyber-focused AI model found thousands of Linux kernel vulnerabilities
Supplied the GPT-6 Cyber naming, the 1,726 variant count and Torvalds’s 8 November comment on kernel fixes.
SUSE to cut about 10% of jobs as CEO cites AI, The Register reports
Established the memo’s wording, the 1,900-employee base, the dates and the EQT ownership and sale background.
Proton apologizes to community after internal memo shows force behind proposed AI use in building its privacy products
Established Proton’s promises, the 1 November manifesto publication and the unclear scope of the apology.
Proton’s AI manifesto and what it means to lead coding agents
Relayed 404 Media’s reporting on the 90% AI-written code figure, the penetration-testing exchange and the tracking-software allegation.
It’s official: Affinity is coming to Linux
Established that Affinity confirmed an in-development Linux version with no date, plus the limits of the Wine route.
Графический редактор Affinity портируют для Linux
Added the same-day iPad launch and how Affinity earns revenue through paid AI features and Canva services.
CutefishOS Is Reborn as Cutefish, Demonstrates Running Mac Apps on Linux
Established the demonstrated apps, the 460,000-line Reef figure, its closed source and the nightly-build status.
New York Times and three other US papers sue Brave over AI scraping
Established the lawsuit’s claims, the five named AI customers, the 30 billion page index and the earlier News Corp dispute.
| Citing this article? Brief excerpts are welcome. Please credit Webiano.digital, name the author where stated, and include a link to https://webiano.digital and to this original article. Full or substantial republication requires prior written permission. Read our Copyright and Content Use Policy. |
This article was prepared with the assistance of artificial intelligence tools. The content underwent expert human review, and Webiano Digital & Marketing Agency assumes editorial responsibility for its final version and publication.
