# AI finds flaws in Windows and sparks a clash with Microsoft

> AI detects flaws in Windows and clashes with Microsoft. We analyze the technical conflict that puts the system's security at risk.
> By José Maldonado · Technology · 2026-06-12

AI has placed Microsoft and Windows at the center of a dispute that could redefine the future of cybersecurity. The ability of new AI systems to detect critical vulnerabilities in millions of lines of code is accelerating the discovery of flaws as never before, but it is also generating growing tensions between large technology companies and the community of independent researchers tasked with auditing their systems.

The confrontation goes far beyond a technical discussion. As AI multiplies the capacity to locate security errors, a debate arises about who should control the discovery, disclosure and correction of these vulnerabilities.

Recent events surrounding the BUMSRAKETE vulnerability in FreeBSD and the friction between Microsoft and certain security researchers illustrate two very different approaches to the same challenge. While some open source projects have opted for more transparent and collaborative disclosure processes, the response of certain large corporations has been perceived by part of the research community as more restrictive and confrontational.

The result is a scenario that transcends Microsoft, Windows or FreeBSD. What is at stake is the model of cooperation that has sustained the security of the internet and critical digital infrastructures for decades. In an era in which artificial intelligence can find vulnerabilities at unprecedented speed, the way companies, developers and researchers manage these discoveries could determine the future of global cybersecurity.

## The «BUMSRAKETE» vulnerability

Recently, it was revealed [**BUMSRAKETE**](https://bumsrake.de/) (CVE-2026-45257), a critical vulnerability in the FreeBSD kernel. This flaw is an example of how the interaction of components that are apparently correct individually can generate serious breaches when combined.

![](https://content.observatorioblockchain.com/wp-content/uploads/2026/06/bumskarete-freebsd.jpg)

But what is it? BUMSRAKETE is a write primitive in the page-cache that allows an unprivileged user to write controlled bytes into the page cache of any file to which they have read access.

How does it work? Basically, the attacker can corrupt executable files, which allows privilege escalation to root (administrative account), at the remote and local level and even with «immutable» files that lose that property when attacked.

**[Vitalik's new plan to improve DeFi security already has prototypes](https://observatorioblockchain.com/en/ethereum/vitalik-prototipos-seguridad-defi/)**

## Impact on infrastructures: The backbone of the network under scrutiny

But the BUMSRAKETE vulnerability (CVE-2026-45257) is not just a technical flaw; it is a reminder of the inherent vulnerability of the fundamental pieces of the Internet. FreeBSD is widely valued for its highly optimized network stack and its ability to manage I/O (input/output) levels extremely efficiently, characteristics that have made it the operating base of critical infrastructures that support much of global traffic.

By allowing arbitrary writing to the page-cache, an attacker could corrupt executables in memory or manipulate data in transit, compromising not only an individual server, but the integrity of the services that depend on it. The real impact of this is measured by the scope of its main users:

*   **Netflix and the Open Connect Network:** Netflix's content delivery infrastructure (CDN) is famous for its use of FreeBSD. Being optimized to send terabytes of video per second, a flaw in the FreeBSD kernel could potentially allow the manipulation of the video data that reaches millions of homes, demonstrating why security at this layer is critical.
*   **NetApp (Storage Systems):** NetApp's ONTAP systems, pillars of corporate storage, use FreeBSD as the basis of their operating system. For a company that handles critical enterprise-level data at petabyte scale, a flaw that compromises the integrity of cached files is an existential threat to information security.
*   **Juniper Networks:** Much of its operating system (Junos OS) is built on FreeBSD. This means that the routing and switching equipment operating at the core of the networks of the world's main Internet service providers (ISPs) depend on the robustness of this kernel.

## Microsoft versus the community: Proactive security or corporate persecution?

The rapid correction of BUMSRAKETE underscores the fundamental difference in how these crises are managed: while in the open ecosystem the priority is technical reporting and transparency to ensure an immediate patch, other platforms, as we will see in the case of Windows, have opted for more complex and less collaborative paths that put the global stability of digital security at risk.

In marked contrast to the maturity and transparency demonstrated in the management of the FreeBSD case, the relationship between Microsoft and the independent researcher known as [**«Nightmare-Eclipse»**](https://www.picussecurity.com/resource/blog/rogueplanet-anatomy-of-the-nightmare-eclipse-microsoft-defender-zero-day) (also identified in some circles as Chaotic Eclipse) has degenerated into a conflict with legal overtones that worries the entire industry.

While in the FreeBSD environment the reporting of BUMSRAKETE followed the channels of responsible disclosure, allowing coordinated remediation and the issuance of patches before mass exploitation, Microsoft's stance has been markedly confrontational.

[Linux's great advantage over Windows: why its security is almost unattainable](https://observatorioblockchain.com/en/technology/seguridad-linux-vs-windows-replicar/)

And it is that, after the revelation of a series of vulnerabilities in Windows Defender (among which BlueHammer, RedSun, UnDefend and the recent RoguePlanet stand out), Redmond's response has not focused exclusively on technical mitigation, but on delegitimizing and pursuing the researcher.

## A dangerous relationship

This dynamic poses dangers that transcend the particular case:

*   **Criminalization vs. Collaboration:** By labeling the findings as «uncoordinated disclosures» and resorting to its digital crimes unit to threaten legal action, Microsoft risks fracturing the security ecosystem. As industry experts point out, when researchers are treated as adversaries instead of allies, the incentive to report flaws ethically disappears, which invariably favors malicious actors operating in the shadows.
*   **The cost to the user:** The conflict has escalated to such a point that the researcher has promised new exploit releases («bone shattering drops»), exacerbating the risk to the systems of millions of users. Microsoft's insistence on the rhetoric of «proactive security» is called into question when patches arrive as a late reaction to public dumpings, instead of being the result of fluid and professional dialogue with the research community.
*   **A dangerous precedent:** This confrontation reflects a deep corporate culture gap. While open source projects are strengthened through external review and radical transparency, Microsoft's strategy toward Nightmare-Eclipse shows that, faced with a critical flaw, the corporate response still prioritizes protecting institutional image over the effective security of the end user.

## Where is cybersecurity heading?

Ultimately, this «trench war» between one of the largest corporations in the world and an individual researcher shows how corporate bureaucracy can take precedence over digital security. If the standard becomes legal persecution, the biggest loser will continue to be the user, whose infrastructure is left exposed while the parties involved waste time in litigation that, by all accounts, should have been avoided through a transparent and humane bug bounty program.

The FreeBSD case shows that transparency and efficient technical resolution are the path to a more robust system. By contrast, Microsoft's response reflects a crisis of trust that can discourage ethical collaboration. For companies and users, the lesson is that in a «vuln-apocalypse» landscape, the choice of platforms should prioritize those that maintain a firm and transparent commitment to security and dialogue with the research community, avoiding being trapped in corporate conflicts that compromise the real protection of data.
