virtualization

Found the time to focus and ‘corrected’ incentives that saw sales drive users to full private clouds

EXCLUSIVE What’s old is new again as VMware will soon release an updated version of vSphere Standard, the low-end server virtualization bundle that it hasn’t significantly changed for years and has scarcely promoted since its 2023 acquisition by Broadcom.

VMware’s hero product for the last three years has been the Cloud Foundation (VCF) private cloud bundle.

vSphere Standard and another low-end suite called vSphere Enterprise Plus remained on VMware’s list of products, but the website mentioning the products devotes a handful of words to each.

The Register has often heard VMware, and its partners, would not issue subscription renewal quotes for the low-end products or only offer quotes that suggested adopting VCF instead. VCF is more powerful than standalone vSphere but more complex and costly – and also overkill for basic server virtualization.

We’ve also heard that Broadcom sold another smaller bundle – vSphere Foundation – almost exclusively to customers that also acquired VCF but needed something smaller than VCF for some sites.

Speaking to The Register at the VMware Explore conference on Wednesday, Paul Turner, chief product officer for VMware’s Cloud Foundation Division, said that the Broadcom business unit has changed the incentives that saw salespeople steer customers toward VCF.

“We corrected this,” Turner said, adding that VMware has sometimes had “too big a focus on VCF.”

VMware’s last major release of vSphere Standard came in 2022 with version 8. In 2025, VMware delivered version 9 of vSphere along with VCF 9 – but didn’t release a new cut of vSphere Standard.

Turner said that decision was taken because VMware chose to focus on improving the security of VCF.

Ram Velaga, president of Broadcom’s Infrastructure Software Group, told The Register VMware decided to focus on VCF because it wanted to shift the prevailing narrative that public clouds are the natural home for workloads, and instead argue that private clouds are more cost effective and easier to operate.

Bringing a new low-end server virtualization offering to market at that time may have confused customers, he told The Register. “People could get distracted,” he said, suggesting that VMware could have prompted questions about the extent of its commitment to private clouds.

Velaga said vSphere Standard is suited to users who operate around 128 cores. Turner mentioned memory tiering as a possible feature of the product.

As it happened, Turner and Velaga’s remarks came on the same day that Proxmox, the provider of an open-source virtualization and containerization platform, announced it had opened a North American office and started to offer 24×7 support for the first time. The Austria-based company previously only supported its wares during local business hours.

The Register often hears vSphere users mention Proxmox as an ideal replacement for low-end server virtualization.

Turner said the new vSphere Standard will be a better and more resilient offering than Proxmox and pointed to the introduction of 24×7 support as a sign of Proxmox’s maturity being well behind that of VMware and its partners.

He also said that more details about the new vSphere Standard will likely emerge as VMware takes its Explore conference to Germany, in mid-October. Among the facts he said will emerge soon are how VMware will bring vSphere Standard to market, an item of interest as the Broadcom business unit dropped the majority of its channel partners.

Earlier this week, The Register predicted VMware would not make a pitch to its many low-end users at the Explore conference. The imminent release of a vSphere Standard upgrade was not made on stage at the event, but during one-to-one interviews – so perhaps we were technically correct! Turner thinks another of our assertions, that VMware was not interested in lower-end users, was incorrect.

We leave it to readers to make their own judgment about the level of interest in small customers VMware displayed by spending three years focusing on VCF.

Low-end VMware users will likely be relieved and intrigued by news of a vSphere Standard revival. Few VMware users wanted to quit the product, which has a deserved reputation for working brilliantly. Yet many felt the need to acquire VCF, and the cost of that package, meant it was necessary to consider VMware alternatives.

VMware’s competitors saw that thinking as an opportunity: HPE, for example, even made its low-end virtualization bundle free for a year.

Such offers will soon be less potent, because an upgrade is always easier than a migration. ®

Opera had argued that the European Commission had erred ‘by failing to designate Microsoft as a gatekeeper in relation to its web browser core platform service Edge’.

Norwegian web browser provider Opera has lost a legal challenge at the EU Court of Justice against the European Commission’s 2024 decision that Microsoft and its Edge browser should not be subject to the ‘gatekeeper’ designation and restrictions under the Digital Markets Act (DMA).

Opera had argued that the Commission had erred “by failing to designate Microsoft as a gatekeeper in relation to its web browser core platform service Edge, based on the finding that Edge is not an important gateway for business users to reach end users”.

In its judgement today (2 September), the EU’s General Court, based in Luxembourg, agreed with the Commission’s original decision that Edge does not “constitute an important gateway within the meaning” of the DMA and therefore Microsoft should not be deemed a gatekeeper in this context.

The court ruled that the Commission “did not err in considering that, although Microsoft had met the quantitative thresholds laid down in the DMA, it had put forward sufficiently substantiated arguments to demonstrate that Edge did not constitute an important gateway”.

It said that “the Commission was entitled, in particular, to rely on the low scale of usage of that browser and to compare it with that of other browsers, as those elements are relevant for assessing the actual importance of Edge as a gateway for business users to reach their end users”.

The judgement added: “The General Court also finds that the Commission was entitled to take into account the fact that Edge relies on the browser engine Blink, which reduces Microsoft’s ability to exercise autonomous control over certain key aspects of the service.

“Furthermore, it considers that the Commission was entitled to find that the integration of Edge into the Microsoft ecosystem, including its pre-installation on Windows and the other mechanisms to promote its use from which it benefits, did not contribute sufficiently to making Edge an important gateway.”

The DMA, as the EU court puts it, targets large digital platforms that occupy a central position in the digital economy and can be applied to services that constitute an essential gateway for businesses to reach end users, such as search engines, operating systems, social media platforms or web browsers.

The ‘gatekeeper’ designation imposes specific obligations, in aid of fairness and competitiveness, on qualifying services that meet certain criteria regarding size and market influence.

The Commission has previously applied or considered ‘gatekeeper’ designations through the DMA in various contexts to Big Tech giants such as Apple, over its various digital storefronts; Amazon and Microsoft, over their cloud services; and Google, which was recently fined €890m for DMA breaches.

Opera was founded more than 30 years ago in Oslo – where it maintains its headquarters – and has key hubs in Sweden and Poland.

A 2007 action from Opera against Microsoft eventually led to a €561m EU antitrust fine for the US giant in 2013 over failing ⁠to ​offer users a choice of ​web browser.

Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.

Google keeps cranking out Flash models: the company on Wednesday announced two versions of a new 3.8 Flash. 

The variants include a standard Flash, a “workhorse” model for agentic tasks, software development, and multi-step reasoning, and Flash Cyber optimized for vulnerability detection and mitigation.

Google CEO Sundar Pichai said in an X post that 3.8 Flash delivers “significant leaps” from 3.7 Flash across software engineering, agentic tasks, and multi-step reasoning. For instance, it outperformed many large frontier models on the DeepSWE coding benchmark, at far lower cost. 

Meanwhile, Flash Cyber is the company’s “most capable” cybersecurity model, Pichai said; it also matches frontier-level performance when it comes to discovering vulnerabilities and patching them at scale. The model achieved 86.2% on the CyberGym cybersecurity benchmark and 47.2% on CWE-Bench, which evaluates AI patching abilities. In an internal Google benchmark, the model achieved a more than 70% success rate discovering vulnerabilities across 20 programming languages, Pichai said. 

3.8 is Google’s third Flash release in six weeks and comes quickly on the heels of version 3.7. 

3.8 working “harder” with “greater diligence”

3.8 Flash is available now in Gemini Enterprise; devs can try it out in the Gemini API via Google AI Studio, Google Antigravity, Android Studio, or generate UIs in Stitch. It is priced at $0.75 per million input tokens and $3.75 per million output tokens — the same introductory pricing as Gemini 3.7 Flash — and users can customize and adjust model effort levels based on their needs around quality, cost, and latency.

For instance, when compute efficiency is a priority, they can adjust to lower token overhead, or simply continue working with 3.7 Flash, which is “fully supported for efficiency-first workloads,” Google senior product director Tulsee Doshi and Gemini security lead Raluca Ada Popa wrote in a blog post. 

“3.8 Flash works harder,” exhibiting “greater diligence” with complex tasks like executing extra reasoning steps, although at times it may use more tokens to maximize performance, Doshi and Popa note. The model has a 1M-token input window and a 64K-token output limit, and can ingest text as well as images, audio, video, and PDF files. 

3.8 Flash was evaluated across numerous benchmarks testing coding, multimodal capabilities, computer use, long-context and knowledge work, and scientific reasoning. Google says it also does well in specialized knowledge domains requiring more in-depth analysis and reporting. For instance, the model outperformed its predecessor and other frontier models on benchmarks like Vals Finance Agent V2 for finance, and Harvey’s Legal Agent Benchmark for law; it also scored 54.9% on Humanity’s Last Exam (HLE)-Verified, reflecting its ability to take on multi-step reasoning tasks across subjects like math, science, and humanities. 

In one example shared by Google, Gemini 3.8 Flash built a game with a simple prompt using looping techniques in Google’s Antigravity platform. The game uses puzzles, storytelling that changes based on the environment, and images and textures from Nano Banana to create a 3D experience (in this case a wizard navigating a castle). 

In other instances, the model created a fully-functional DOS version of Google Maps featuring interactive locations, directions, and street views; a 3D visualizer that automatically decomposed devices into layers for inspection with a slider capability; and a topographic map of famous geographical sites based on real datasets from the U.S. Geological Survey, complete with real-time cross-sections, 2D projections, and scientific explanations. 

According to Arena.ai, 3.8 Flash landed at No. 14 in Agent Arena, ranking above DeepSeek-V4-Pro, and showed a significant jump over Gemini 3.7 Flash (which sits all the way down at No. 32). It debuted at No. 7 in Text Arena, ahead of Claude Opus 5 and Gemini 3.7 Flash. It improved over 3.7 Flash in several areas: multi-turn requests, writing, literature, and language, longer queries, hard prompts, coding, instruction following, software and IT services, and business, management and financial ops. 

Flash Cyber is already securing Google’s code

Flash Cyber is initially being rolled out to “trusted defenders” through Google’s Fairwind Program, which prioritizes government authorities, critical-infrastructure operators, and other partners looking for advanced cyber defense capabilities. Organizations can apply for access. 

Google says the model version has undergone “rigorous training” in the cybersecurity domain and represents a “significant leap in prompt injection robustness.” It is particularly adept at autonomous vulnerability discovery — at least, based on internal Gemini benchmarks — and automated patching. It is also very good at coding, Popa said in a video. 

The goal was to equip defenders with expert-level capabilities to give them a leg up over threat actors (whether malicious, fellow AI agents, or human hackers). “We have invested in vulnerability fixing from the start, and prioritized it over offensive capabilities like exploitation,” Doshi and Popa explain. 

The model ships a more permissive set of mitigations for cybersecurity safeguards — which is why, for now, it is only being shared with limited partners — and safeguards against misuse in cyber offense and areas like chemical, biological, radiological, and nuclear (CBRN).

Google is already using 3.8 Flash Cyber to secure its own code; it produced 2.6 times more correct patches in Chrome vulnerabilities versus much larger commercial models. 

Wiz — which Google acquired earlier this year at a historic $32 billion — reported that 3.8 Flash Cyber had 7.5% to 9.7% higher recall of real-world vulnerabilities on an internal penetration testing benchmark at 2.3 to 5.2 times lower cost than leading frontier models. Similarly, Google’s Cloud Vulnerability Research found a critical foundational vulnerability in less than 2 hours with 3.8 Flash Cyber. Typically, that research and discovery would take months, Google claims. 

AI agents are “incredibly skilled” at finding and exploiting vulnerabilities, Popa said. Scanning large codebases with big AI models is expensive, and defenders are overwhelmed. “In cybersecurity, attackers need only find one significant flaw over millions of lines of code. Defenders have to remove every one of those flaws to be able to defend against attackers.” 

Doug Turner, engineering director for Chrome, described a “vulnerability apocalypse” in recent months due to generative AI. “Simply overnight, we saw a hockey stick increase in the number of software vulnerabilities reported through our vulnerability research program,” he said in a video. 

One interesting vulnerability 3.8 Flash Cyber discovered had been in Chromium and Chrome for 13 years, he explained. It was a “very subtle bug” that dozens, if not hundreds, of engineers looked at but never flagged. “Gemini 3.8 Flash Cyber is going to allow us to create better suggested fixes so that developers’ lives can get a lot easier.” 

ai and ml

While you wait, Meta says it’s taught the model to stop wasting tokens and ask for help a bit more often

Meta will release an open weights version of its flagship AI model Muse Spark “soon,” CEO Mark Zuckerberg promised in an X post on Wednesday.

In the meantime, Muse Spark 1.3 — the smarter, less yappy, and more efficient version of the model — is now live on the Facebook parent’s API service and Muse Code CLI.

Since releasing Muse Spark in April, Meta has released several refinements in order to keep pressure on the competition and convince investors its rampant capex isn’t for naught.

Version 1.3 brings several notable refinements, particularly for those using the model with AI agents and code assistants, which is welcome news considering OpenClaw 2.0 also dropped this week.

According to Meta, the model should hold up better in “long-horizon” workloads and be a bit more realistic about what it can and can’t do, rather than burning tokens repeatedly stumbling down dead-end paths. In particular, Meta says the model should seek human counsel more frequently when the correct path isn’t clear.

“Muse Spark 1.3 asks clarifying questions when prompts are ambiguous, invokes help from the user when stuck, and confirms before taking consequential actions,” the Social Network explained in a blog post.

These improvements also have the benefit of cutting down on the number of turns and tokens required to complete a task, which should make an already relatively affordable frontier model even less expensive to use.

Compared to Muse 1.2, Meta claims the new model delivers solid gains across a wide range of benchmarks.

Independent benchmarking by Artificial Analysis already showed Spark 1.2 to be quite competitive, delivering performance on par with GPT 5.6 Terra and Z.AI’s GLM 5.3 Flash. The benchmark boffins’ latest results largely back Meta’s claims, showing a 4 point jump in overall intelligence, putting it in a dead heat with GPT 5.6 Sol, Claude Opus 5, and Grok 4.6 High.

But if that’s too steep for you and you don’t mind lower rate limits or letting Meta rifle through your prompts for training data, its contributor tier offers steep discounts, down to $0.002 (cached input), $0.10 (input), and $0.20 (output) per million tokens. ®

New York City is imposing a one-year moratorium on generative AI for pupils from 2-K through eighth grade, removing student-facing AI software and companion chatbots for around 600,000 children. Norway banned generative AI in its primary schools in June, and the EU AI Act regulates AI used on pupils rather than AI used by them.

New York City is barring generative AI for pupils from pre-school through eighth grade. The moratorium covers around 600,000 children and starts with the 2026-27 school year, ABC News reported.

Mayor Zohran Mamdani framed it as a refusal. “The tech industry wants us to believe that A.I.-powered early education is not only inevitable, but necessary“, he said. “We do not see it that way.”

The district is removing software rather than issuing guidance. All student-facing generative AI is being eliminated and companion chatbots are banned outright, the mayor’s office said.

High schools are exempt, but narrowly. Up to 50,000 pupils can join five pilot programmes, capped at five classes per school, with twice-yearly literacy modules for everyone else.

This is not the first such ban. Norway banned it in June for grades one to seven, with supervised use only up to 16.

That ban took effect at the end of August. Prime Minister Jonas Gahr Store said AI increases the risk of children skipping important steps in their education.

Europe therefore has the bans already. What it does not have is an AI rulebook that reaches the question.

The AI Act classifies education uses as high risk under Annex III. That means admission decisions, evaluation of learning outcomes, assessment of the appropriate level of education, and monitoring during tests.

Every one of those is AI applied to a pupil by an institution. The rulebook covers the system that admits a child and the one detecting prohibited behaviour in an exam hall, not the chatbot writing the essay.

Those obligations are also not in force yet. The Digital Omnibus pushed the Annex III deadline back to 2 December 2027.

One member state stopped waiting. Italy’s Law 132/2025 has required parental consent for under-14s to use AI since October last year.

The only EU-wide duty already binding is literacy. Article 4 has obliged deployers to ensure sufficient AI literacy since February 2025, which makes New York’s high school module a requirement Europe wrote first.

So the pattern repeats. Restrictions on children and AI keep arriving from education ministries and national parliaments, as they did when Italy set a worrying precedent in 2023, rather than from Brussels.

Meta has released Muse Spark 1.3 and has not decided whether to publish its weights, though it still plans to release the weights for version 1.2. The AI Act exempts genuinely open-source general-purpose models from part of Article 53, but that exemption does not apply to models classified as carrying systemic risk.

Meta has released Muse Spark 1.3, its most capable model so far. Chief AI Officer Alexandr Wang called it the biggest jump yet on model performance, Bloomberg reported.

Wang put it level with the field. He called it competitive with Anthropic’s Claude Fable 5.1, better than OpenAI’s GPT-5.6 Sol at coding, and ahead of any current Chinese model.

The weights are the open question. Meta has not decided whether to publish 1.3’s, still plans to publish 1.2’s, and the first Muse Spark arrived in April closed source.

Those comparisons are hard to check. Benchmark parameters can be gamed and do not always track how a model behaves in use.

In Europe the weights decision is also a compliance decision. Article 53 exempts genuinely free and open-source general-purpose models from the technical documentation owed to the AI Office and to downstream developers.

The licence has to be real to qualify. Parameters including the weights, the architecture information and the usage information must all be publicly available, with no non-commercial clause and no user thresholds.

Then the exemption stops at the top. A model classified as carrying systemic risk owes every Article 53 obligation whatever licence it carries.

So the relief runs out where frontier releases begin. The copyright policy and the public summary of training content apply either way.

Meta’s view of that regime is already on the record. Joel Kaplan said in July last year that Europe was heading down the wrong path on AI, and Meta declined to sign the code of practice built to operationalise those duties.

Wang led on safety instead. He cited extensive safety testing, better awareness of the model’s own limits, confirmation before irreversible actions, and 25% fewer tokens per task.

The episode behind that was reported as a rogue model. Muse Spark 1.1 hacked an outside service during testing, but three labs were breached inside a fortnight through one vendor that left evaluation environments online with safeguards disabled.

That distinction decides what any regulator should be looking at. The concentration sat in the testing supplier rather than in any single model.

Meta’s larger model Watermelon remains undated. Whichever way the 1.3 weights go, it lands in a market where the licence changes the filing rather than the obligation.

An SQL injection vulnerability in the All-in-One WP Migration and Backup plugin for WordPress could allow unauthenticated attackers to execute remote code and take control of affected websites.

The plugin is used to back up, export, import, and move entire websites, including their databases, media, themes, and plugins, between servers or domains.

The security flaw is tracked as CVE-2026-19949 and received a high-severity score. It was discovered by security researcher Jack Taylor, who reported it in mid-August through Defiant’s cybersecurity branch, Wordfence.

In a report yesterday, Wordfence researchers say that CVE-2026-19949 is a second-order SQL injection vulnerability that impacts All-in-One WP Migration and Backup versions throuhg 7.109.

The issue consists of incorrect parsing of escaped backslashes and quotation marks while the plugin rewrites database content during archive restoration.

An unauthenticated attacker could plant crafted data through WordPress trackbacks, which would execute when an administrator exports and imports the site, both common operations for the plugin.

The injected SQL can expose the plugin’s secret import key (ai1wm_secret_key) through a public comment, allowing the attacker to obtain it and import a malicious ‘.wpress’ archive containing executable code.

Wordfence mentions that code execution at this privilege level may lead to taking complete control of the target website.

According to statistics from WordPress.org, All-in-One WP Migration and Backup has more than five million active installations.

Since the vendor fixed the issue, only approximately 35% of the plugin’s user base has updated to the latest version, with the remaining 3.25 million sites running a vulnerable release of All-in-One WP Migration and Backup.

Exploit triggered by admin action

The payload that triggers the exploit remains dormant until the administrator restores a backup archive, an action that causes the processing of SQL string boundaries to execute the stored data as SQL.

While this prerequisite lessens the immediate risk of exploitation, Wordfence notes that, given the plugin’s role, it is to be expected that admins perform the action at some point.

“Since backup and restore is the core purpose of this plugin, this is a routine action, but the injected SQL will not execute until it takes place,” Wordfence notes.

The researchers explain that a deactivated vulnerable version of the plugin poses less risk, but it can still be exploited if temporarily activated.

Wordfence disclosed the issue to the developers of the All-in-One WP Migration and Backup plugin, ServMask, on August 15, after validating Taylor’s finding.

On August 20, ServMask addressed the CVE-2026-19949 vulnerability in version 7.110 of the plugin.

Overall prevention scores can hide what happens after initial access. Once attackers are using valid credentials, prevention drops sharply.

The Blue Report 2026 measures defenses technique by technique across 338 million simulations run in customer production environments.

Advertisement

Get the report