Skip to content
PodSubs
ShowsPeopleGuestsJobsNewsEvents
More
BlogPodCred
Menu
ShowsPeopleGuestsJobsNewsEventsBlogPodCredNetworksSign in
Join nowSign in
  • Shows
  • People
  • Jobs
  • News
  • More
    • Guests
    • Events
    • Blog
    • PodCred
    • Networks
    Join nowSign in

Tech Talks With Kinsoft

Tech Talks With Kinsoft artwork

Last Week in Tech - A Brutal AI-Driven Tech Sell-Off, a Perfect-10 Siemens Flaw Under Attack, and Cisco's Exploited VPN Bug

Nov 8, 202621 minAudio
ShareXLinkedInFacebookWhatsAppEmail

Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! Last week in tech, 30 Oct to 6 Nov 2026. US tech stocks sold off hard on Thursday 5 November: the Nasdaq Composite fell 5.52%, the S&P 500 4.36% and the Dow 3.44% at the close, with the semiconductor index down about 12%. Reported causes include doubts over AI spending returns, profit-taking and midterm-related worries about data-centre limits; these are reported explanations, not confirmed causes. The US added 165,000 jobs in October, with unemployment at 4.1% (BLS). OpenAI says GPT-6 Astra was its first model rated Critical for cyber capability; GPT-6 Cyber is reportedly also Critical (single trade-press source). Meta bought AI coding startup Conductor; its cloud service shuts 20 November. Security: Siemens SIMATIC IoT2050 Node-RED flaw CVE-2026-58115, CVSS 10.0 (v3.1 and v4.0), exploitation reported by VulnCheck from 30 October. Cisco ASA/FTD CVE-2026-76418, CVSS 3.1 score 8.6, exploited denial of service, added to CISA KEV 29 October. Visit www.kinsoft.com.au to talk through your security and IT needs. Sources: BLS; Fox Business; Yahoo Finance; IBTimes; Investing.com; OpenAI; Techzine; TechCrunch; Siemens ProductCERT SSA-834709; The Hacker News; Help Net Security; Cisco; Rescana.

Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.

Transcript

From the show. Tap a time to play from there.

Imagine a multi -billion dollar factory floor. It's, you know, humming with robotic assembly lines, perfectly calibrated to produce physical goods around the clock. Right. Just a highly tuned physical operation. Exactly. Now imagine the digital front door to that factory just vanished. Like no locks, no hinges, just wide open to the public internet, waiting for literally anyone to walk in. Which is a terrifying thought. It really is. And today... We are looking at how

a massive financial sell -off on Wall Street is, believe it or not, fundamentally tied to that exact missing door. Yeah, we often treat technology news as these distinct, isolated categories, like markets do one thing, artificial intelligence labs do another, and cybersecurity sits in its own little silo. Right. But the reality is that it's all an ecosystem. It's an ecosystem where financial pressure at the top forces rapid... technical deployment at the bottom. And that, well, that eventually breaks the physical hardware

we rely on every single day. Yeah. And that causality is the entire mission of our deep dive today. We're pulling from a really dense, highly detailed technology and security briefing. It was originally sourced from a November 9, 2026 report by Kinsoft. The last week in tech report, right? Yep. That's the one. And our goal for you, the listener, is to basically synthesize this Massive ripple effect. We are tracking the chain reaction from macroeconomic forces trickling all the way down

into the deployment of advanced AI coding agents. And landing squarely on some highly critical, actively exploited cyber vulnerabilities in industrial control devices. Exactly. Okay, let's unpack this. And for those of you watching the video feed, you'll actually see my backdrop transitioning into this dynamic, interconnected web of data streams and financial ticker symbols. Oh, yeah. It looks really cool. Thanks. We're doing this to map it out step by step. Because seeing the macroeconomics intersect directly with the underlying technology, I mean, it really is the only way

to understand what is happening right now. Yeah. So let's start at the very top of that web, right? Just to understand the financial pressures driving current tech behavior. Good place to start. So on Thursday, November 5th, 2026, U .S. tech stocks suffered a profoundly heavy fall. I mean, we are talking about the NASDAQ composite dropping 5 .52%. Which is huge. Huge. The S &P 500 down 4 .36%. And the Dow falling 3 .44%. But the bleeding

was actually most severe in the specialized hardware sectors. Right, the AI hardware guys. Yeah. NVIDIA dropped roughly 9%. And the Philadelphia Semiconductor Index plummeted an astonishing 12 percent in a single session. Yeah. And a 12 percent drop in a sector index in one day. It really requires us to look at the mechanics of modern institutional trading. Because it wasn't just guys in suits yelling sell on a trading floor. No, not at all. This wasn't a bunch of human brokers slowly deciding to sell. We're talking about algorithmic risk

models hitting their maximum pain thresholds. Wow. Yeah. Once initial selling begins, it drains the liquidity in the market. The algorithms see the price dropping. They mathematically recalculate their risk profile of these massive AI hardware investments. And they automatically trigger cascading sell orders before a human even steps in to intervene. It's just machines reacting to machines. Exactly. Well, the Kinsoft report actually highlights three specific catalysts that triggered those algorithms. And to be totally clear to you listening, these are reported market anxieties. They are

not verified mathematical constants. Right. It's what the market is feeling. Yeah. The vibes, essentially. Yeah. So first, institutional investors are developing some serious doubts about when all this massive capital expenditure in AI is actually going to yield a tangible return on investment. The ROI problem. Right. Second, there is simple profit taking. Funds are just cashing out their chips on this year's best performing stocks. Makes sense. And third, there were genuine worries that the U .S. midterm election results could lead to new regulatory restrictions on data center construction. Yeah, the energy and

infrastructure regulations. Right. And we're looking at that last point purely impartially as a market fear. We're not taking any political stance here. But, you know, the market priced in that anxiety regardless. Absolutely. And the financial firm Jefferies described this violent movement as, quote, reactionary rather than a fundamental reset. Which is a bit of a relief. It is. And the data actually backs that up. Friday saw a partial bounce in chip stocks, though the report does note there was no verified closing figure at the time of publication. Right. But if we connect this to the bigger picture, this

market anxiety over AI's return on investment sets the absolute baseline for how enterprise technology is going to operate in the coming year. Yeah. And the business lesson for you, whether you are running an IT department or leading corporate strategy, is critical. The report explicitly warns that if your IT roadmap leans heavily on a single AI vendor's pricing model, you must build in flexibility for that pricing to fluctuate wildly. Because when these AI vendors feel the squeeze from Wall Street investors demanding immediate returns, I mean, they have to make

up their profit margins somewhere. Right. And unfortunately, the end user is always the final shock absorber for corporate debt. It's so true. Think about it like buying a state -of -the -art luxury vehicle for your company fleet, but then realizing, after you sign all the paperwork, that the engine only accepts a highly proprietary type of fuel. Oh, that's a great analogy. And you can only purchase it from one specific highly volatile gas station. If the owner of that station decides to quadruple the price overnight because their own investors are panicked, you are entirely at their mercy. You're just stuck. Totally stuck.

You have no alternatives. You need to design your IT budget with shock absorbers to handle those venture price hikes. Yeah, and that actually leads directly into the second massive pressure point we need to talk about. Okay, let's hear it. We mentioned the market sphere of regulatory restrictions on building data centers, but there is a much heavier force impacting the physical infrastructure of the AI boom, and it originates from a seemingly unrelated macroeconomic metric. Which is? The broader labor market. Okay, yeah.

According to the Bureau of Labor Statistics, the U .S. added 165 ,000 jobs in October. Right. Bringing unemployment down to a really firm 4 .1%. And the report notes that economic commentators suggest this firm labor market will keep interest rates higher for longer. Yes. And this is where I actually have to pause and ask you something. Wait. Normally a strong jobs report is a good thing. Why would adding 165 ,000 jobs secretly be bad news for tech infrastructure? Right. It seems entirely contradictory, at least until

you look at the macroeconomic chain reaction. Okay. Walk me through it. Well, the report clarifies this is market commentary, not official Federal Reserve policy. But the logic holds up perfectly. A surprisingly strong jobs report signals a firm labor market. Right. People are working. Exactly. Meaning consumers have disposable income. That consumer spending keeps inflation stubborn. It stays sticky, as they say. And when inflation refuses to cool down, central banks are basically forced to maintain high interest rates. And high

interest rates are the natural enemy of capital expenditure. They absolutely are. Look, building a modern AI data center is not just pushing a software update. It requires billions of dollars in physical capital. We are talking about land, massive cooling towers, power substations, and racks upon racks of specialized silicon. It's heavy industry, basically. It really is. Tech giants finance these projects by borrowing capital. So when interest rates are kept high due to that stubborn inflation, the cost to service that

debt just skyrockets. It suddenly becomes incredibly expensive to finance the physical foundation required to train the next generation of AI. So because borrowing money to build data centers is suddenly so crazy expensive, the AI labs are desperate. They need to prove their models are worth the massive premium price tag to their investors. Exactly. The pressure is immense. And that desperation is pushing them to release highly specialized, just aggressively powerful tools. The capabilities are escalating rapidly

and specifically in the highest stakes arena possible, which is cybersecurity. Yeah, we are seeing the model shift from generalized chat assistance to highly specialized operators. OpenAI actually stated that its GPT -6 Astra model, which was announced back in September, that it was its very first model to reach the, quote, critical level. for cybersecurity capability under its internal preparedness framework. And just for some context for you listening, that preparedness framework is their internal rubric.

It tracks how capable a model is at offensive cyber operations, like finding and exploiting zero day vulnerabilities or how powerful it is defensively. So hitting that critical threshold. means the model possesses highly specialized cognitive capabilities in network architecture and code analysis. And the escalation continues from there. The Kinsoft report details this rumor regarding a model called GPT -6 Cyber. Oh yeah, I saw that. It is reportedly available only to

select vetted customers and is also rated at the critical level. Now, the report emphasizes this detail stems from a single trade outlet, so it should be treated strictly as unverified reporting at this stage. Right. We have to take it with a grain of salt. Exactly. Yeah. But the corporate behavior surrounding these capabilities, that is undeniable. Yeah. You can see that corporate behavior directly in Meta's latest move. They just confirmed the acquisition of a startup called Conductor. Right. And Conductor is a company that specializes in running several AI coding

agents in parallel. The financial terms of the acquisition were undisclosed, obviously, but Conductor's public cloud service will be permanently shut down on November 20. Yeah, buying a startup just to shut down its public -facing service less than two weeks later is a pure absorption play. Totally. Meta clearly wants the underlying architecture and the engineering talent entirely isolated for their own internal infrastructure. Here's where it gets really interesting, though. We need to talk about what running... quote, several AI coding agents in parallel actually

entails. It's wild. It is essentially instantly cloning a team of junior software developers. Instead of one AI writing a piece of code linearly from top to bottom, you have a swarm. A swarm of AI. Yeah, but wait. If multiple AI agents are coding in parallel simultaneously, how do they avoid creating a fragmented, Frankenstein -like code base? Like, who is the digital traffic cop? compiling all of this what's fascinating here is the mechanism behind it you typically

have one supervisory agent that takes the human prompt and breaks it down into modular micro tasks okay the parallel agents then write those individual modules simultaneously but and this is the key part before anything is compiled the agents actively cross -check each other's syntax against a unified test suite oh wow yeah they actively look for memory leaks or logical errors in their peers code it is massive parallel cognitive processing tailored specifically for software development and debugging. And what makes that

worse, honestly, is the dual -edged nature of that capability. Right. If an AI system can analyze a massive complex code base at a critical level to debug it, it possesses the exact same mechanical capability to analyze a code base at a critical level to break it. Yep. The intelligence required to patch a system is essentially indistinguishable from the intelligence required to exploit it. So the AI models are reaching escape velocity in their software capabilities, but the actual

cybersecurity battleground on the ground, the vulnerable physical hardware running our industries, our pipelines, and our networks, is still stuck in the mud. And the friction between advanced AI and legacy hardware is where the damage actually happens. Yeah. So let's look at the real -world targets that are failing to keep up. The report highlighted two massive vulnerabilities at the network edge this week. Both very serious. The first involves Siemens, specifically vulnerability CVE -2026 -58115. This affects Siemens' SIMATIC

IoT 2050 devices that are running industrial OS with Node -RED. Right, and we really need to define Node -RED here to understand the severity. Go for it. Node -RED is a flow -based programming tool. It basically gives industrial engineers a visual web interface to wire together hardware devices, APIs, and online services. Okay. It's the connective tissue that allows a physical machine on a factory floor to send operational data to a cloud server. Got it. And Siemens gave this vulnerability a maximum severity score of

10 .0 under both CVSS version 3 .1 and the newer version 4 .0. A perfect 10. Yeah, a perfect 10. And for you listening, a 10 .0 in the common vulnerability scoring system means the building is structurally compromised. It is highly exploitable, works remotely over a network, and requires absolutely zero user interaction or special privileges to execute. Which is worst case scenario. Seriously. The specific flaw is missing authentication, which allows for remote command execution. Yeah,

let's translate missing authentication for a second. It does not mean the attacker is using advanced AI to guess a complex cryptographic key or bypass a sophisticated firewall. It means the system never even asks for an ID in the first place. Returning to your factory analogy from earlier, the digital front door doesn't exist. Just wide open. Exactly. Because this flaw is in the Node -RED interface, an unauthenticated attacker can literally just walk in and start issuing commands to manipulate every physical machine that specific node talks to. That is

wild. Siemens published the advisory for this back on August 11, and then they updated it recently on November 4. And threat actors have... already found the missing door. Of course they have. Yeah. Volncheck reports that in the wild, exploitation of this vulnerability actually began on October 30. Wow. Though, the report does note the CISA catalog status was unconfirmed for this specific exploit at the time of writing. Now, the flaw is fixed in versions from v4 .3 .4 -1, but the strict underlying advice in the report is incredibly

simple. These devices should not face the open internet. Which, you know, brings up a massive tension in modern industry. What do you mean? Well, companies want to monitor their factory floors or pipelines or assembly lines in real time from a headquarters that might be halfway across the world. Right. They want the data. Exactly. They want to pipe operational technology data directly into cloud analytics platforms to optimize efficiency. So someone on the ground just plugs an industrial edge device directly into the public network to make the data flow faster. Entirely bypassing proper network segmentation.

Yes. They prioritize immediate business analytics over fundamental security hygiene. With a terrifying 10 .0 severity score, why on earth are industrial edge devices like the Siemens gear ever connected directly to the public internet in the first place? Like, they trade basic network isolation for a faster conveyor belt. And it is a catastrophic trade. Truly. But Siemens wasn't the only company dealing with a major architectural fire this week. The Kinsoft report also detailed a critical issue for Cisco. Right. The VPN issue. Yeah.

Vulnerability CVE -2026 -76 -418. This flaw sits in the Cisco ASA and Firepower threat defense systems, specifically targeting the remote access SSL VPN service. We are talking about the primary gateways designed to let remote employees securely access the internal corporate network. And the CVSS version 3 .1 score on this one is 8 .6, marking it as high severity. And the mechanism of the exploit is what makes it so disruptive. An unauthenticated attacker can force a device reload, which essentially causes a denial of

service. Yeah. But how does an outside user force a corporate firewall to just reboot itself? It's actually a pretty brutal technique. When a remote worker tries to connect to the VPN, there is an initial SSL handshake to establish the secure tunnel. Right, the standard protocol. Yeah. But the attacker isn't even attempting to guess a password here. They are sending a specifically crafted malformed packet during that very first handshake. The firewall's operating system tries to parse the garbage data. It allocates way too much memory trying to figure it out. And it just

exhausts its internal buffer. So to prevent a total system collapse, the firewall intentionally panics and forces a reboot. Oh, wow. And when the firewall reboots, it drops every single active VPN tunnel. Yep. Every single one. And in 2026, I mean, hybrid work relies entirely on these remote access VPNs. Taking down the VPN isn't just a technical glitch. It literally severs the workforce from the network. It shuts down business operations completely. And Cisco confirmed this vulnerability has absolutely been exploded in the wild. They published the advisory on October

28th. CISA, the Cybersecurity and Infrastructure Security Agency, added it to their known exploited vulnerabilities list the very next day on October 29th. That was fast. Very fast. And they assigned a strict federal patching deadline of November 1st. Wow. Yeah, that aggressively tight turnaround for a federal deadline illustrates exactly how crippling an actively exploited VPN vulnerability is to national infrastructure. For sure. The report does mention a workaround. Administrators can disable secondary authentication on the affected

tunnel group. But the report is super emphatic that patching is the only real answer here. Yeah, relying on a workaround is like putting a bucket under a leaking pipe. Deploying the patch is actually shutting off the water and fixing the plumbing. That's a great way to put it. Thanks. What is most striking, though, when you look at the entire board is the intense friction. The friction. Yeah, we have a technology ecosystem trying to move at light speed. deploying parallel AI coding agents and financing massive data centers.

But it is constantly dragging against the friction of physical reality. High interest rates making debt expensive, vendor lock -in risks, and decades -old security flaws like missing authentication on internet -facing devices. So what does this all mean? The Kinsoft report distills this massive web down to a very specific visceral takeaway. They targeted Australian organizations in their briefing. But honestly, the rule applies globally. Absolutely. If you are listening to this on a

Tuesday and you don't know the exact name of the engineer who is contractually obligated to patch your Siemens factory floor gear or update your Cisco VPN, you are currently operating on borrowed time. Accountability is the ultimate security patch. Wow. Yeah. If you have an industrial device wired into your network and no specific human being has it in their active job description to monitor its firmware, that device is not an asset. It is an open door for an attacker. You cannot secure what you do not know you have.

And you cannot maintain it if nobody owns the process. Exactly. If you want to dive deeper into how this applies to your own operations and architecture, the source of today's insights explicitly invites you to visit www .kinsoft .com .au to talk through your security and IT needs. They are the ones who pulled these vital threads together for this week's report. Tracing this ripple effect leaves me with one final, deeply provocative thought to mull over, though. Oh, lay it on me. Well, we look at the macro picture, right? AI models like OpenAI's Astra

are reaching critical offensive and defensive capabilities at the exact same time our physical infrastructure is being hammered. by fundamental vulnerabilities. Yeah, the timing is wild. Yeah. It makes you wonder, as human security teams struggle to keep pace with these exploits, are we rapidly approaching a future where the only thing fast enough to patch our networks and defend our infrastructure is the very same artificial intelligence that threat actors are using to attack them? Oh, man. An automated war playing out in milliseconds, right beneath our feet.

We are literally just trying to keep the VPN online. Precisely. Well, that's a thought to keep you up at night. Keep questioning the data around you. Keep looking for the underlying mechanisms driving these ripple effects. And we will catch you on the next deep dive.

Open the transcript file

Hosts and Guests

Nobody is credited on this episode yet.

Credit the people on this episode

Be the first to rate this episode

More episodes

All episodes
  1. Iris Ohyama - A Suspicious File in the Cloud, 7.7 Million Customers at Risk, and an Online Store That Went Dark

    Nov 19, 202620 minAudio

    Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! Japanese home goods company Iris Ohyama (online store operator Iris Plaza) found a suspicious file in an external...

    Credit the people on this episode

  2. Who Gives A Crap - A Marketing Supplier Breached, Six Days to Tell the Client, and Customer Data That Didn't Need to Be There

    Nov 17, 202619 minAudio

    Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! Australian online retailer Who Gives A Crap told customers on 7 November 2026 that their information was accessed...

    Credit the people on this episode

  3. Last Week in Tech - An Outlook Zero-Day That Leaks Your Password Hash, Chrome's Eighth Exploited Bug, and Nvidia Without Jensen Huang

    Nov 15, 202621 minAudio

    Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! Last week in tech, 7 to 14 November 2026. Microsoft's November Patch Tuesday (10 November) fixes well over a...

    Credit the people on this episode

  4. Chainalysis - From "Non-Sensitive" to "All Users": How a Breach Notice Changed in Five Days

    Nov 12, 202619 minAudio

    Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! Chainalysis disclosed on 30 October 2026 that an external actor using "offensive agentic techniques" accessed...

    Credit the people on this episode

  5. NSW Health - A Dental Voucher Website, Four Gigabytes Taken, and the Providers Whose Details Were Exposed

    Nov 10, 202623 minAudio

    Got something to say to this podcast? Send a voicemail, and you might hear yourself on the next episode! NSW Health reported unauthorised access to its Oral Health Fee For Service Scheme website on 29 October 2026;...

    Credit the people on this episode