The Weekly Cybers #143
Everything is rubbish and falling apart, Jensen Huang is unlikely to return to Nvidia, Anthropic is history’s biggest share float, and much more.
20 November 2026
Welcome to a collapsing world
Well then. A lot of things are failing. Maybe it’s a coincidence. Or maybe it’s a symptom of systemic changes in the technology industries. Who can say? We’ll see a lot of discussion in the coming weeks.
It’s a huge week in tech policy news, and I'm sure I’ve only scratched the surface. Some of these stories are also from last week because I had to cut things short. Here we go.
Yes, everything is rubbish and is falling apart
So last week we had the global Microsoft outage, which we have since discovered was caused by a bug introduced three months ago. There were network failures at the NSA, the US National Security Agency. America also had air traffic control failures. And there was much more, including some banks.
Donald Trump says “relax”, so yes everything is OK, right? Is it just coincidence? That everything is just more complicated now?
On Thursday I was recording a podcast with John Birmingham because he’s got a new Quarterly Essay out, Another Country The Rise and Rise of One Nation. He’s got a theory which he outlined in his own weekly essay, AlienSideBoob, and which makes a lot of sense: businesses big and small piled into AI without really knowing whether it would work or not.
“For two years now, the corporate world has been looking for efficiencies and advantages, which often means sacking human beings and replacing them with software. Or sometimes sacking expensive human beings, like 45-year-old senior programmers, and replacing them with recent graduates, who are then ordered to vibe code software to add to the company’s code base. All of which has come apart spectacularly in the last week or so, it seems, as these systems begin to buckle and collapse under the accumulated errors introduced over the past two years,” he writes.
“We’ll be a while figuring out exactly why so many IT systems failed one after the other but one whimsical explanation I quite liked was that in the US, companies like to get their software patched and updated before Thanksgiving... And once the failures started, they cascaded.”
This is not to say that AI is always bad. The highly structured nature of program code means that it’s great at finding bugs as well as creating them.
Cybersecurity startup Calif has used AI to find a critical bug in Java, the world’s third most used programming language. The same company found the critical bug in PHP that caused a bit of stress last month. There will be many more to come — although the finding can be done by The Bad Guys™ as well as the defenders.
The Australian Cyber Security Centre (ACSC) has repeated its warning about the growing pace of these disclosures. Microsoft boss Satya Nadella has also warned that AI makes outages more likely.
So will businesses and indeed, governments now take the time to improve the stability of their systems before chasing The Next Big Thing?
Jensen Huang has fans, but he’s not returning
The fallout continues after the departure of Jensen Huang from his job as chief executive of chipmaker Nvidia.
Nvidia is facing an enormous push to restore Huang, the world’s 13th richest person. Support is coming, inevitably, from Elon Musk and Donald Trump, but also Epic Games CEO Tim Sweeney, a bunch of prominent shareholders, and a petition with 180,200 signatures.
“Taiwan’s tallest skyscraper, Taipei 101, was illuminated on Monday with messages including ‘Taiwan Loves Jensen’,” reports Reuters.
But Nvidia says no, as do corporate governance experts and the US treasury secretary. At least not until the investigations are complete.
Oh, and Nvidia could well have been hacked, following claims by ransomware group The Gentlemen’s affiliates. Something else to think about.
PLEASE CONSIDER SUPPORTING THIS NEWSLETTER: Unlike Jensen Huang, I am not a billionaire. The Weekly Cybers, as well as my podcasts The 9pm Edict and Another Untitled Music Podcast, are supported by their audiences. You’re more than welcome to throw in a tip. If you’ve already done so, thank you.
Also in the news this week (and from last week)
- Sensitive data from Centrelink and other Australian agencies may have been fed to Palantir software via the Australian Criminal Intelligence Commission (ACIC) as part of an anti-fraud program.
- The City of Sydney wants to install AI-based CCTV scanners around the inner city. According to Sydney Criminal Lawyers, “The proposed updated code stipulates that the biometric facial recognition tech used to automatically recognise human features in public is banned. Rather, the AI tech being employed can scan the public in real time to identify specific kinetic movements that suggest criminal behaviour may be about to happen. This can involve groups gathering or vehicles going in the wrong direction.”
- Research on the most popular chatbots shows that when they’re asked about Australia government services just one in five answers are reliable. I wonder what the figure would be for asking random humans for advice.
- “[The] Rapid Review into Australian Government arrangements for an AI-driven cyber incident examines the OpenAI cyber incident affecting Services Australia,” and we now have its final report. However we did write about this last week, based on the press releases.
- Australia has signed on to a US-led AI supply chain agreement, which is all about competition with China.
- The Australian music industry doesn’t like the AI copyright proposal that’s reportedly in the works.
- ANZ Banking Group says it used AI to shift a 1.7 petabyte data platform to Amazon Web Services in one-third of the expected time, six months rather than 18 months. So yeah, sometimes it does work.
- Datacentre opposition has gone from marginal to mainstream in five months. You may have noticed.
- The Independent National Security Legislation Monitor (INSLM) has issues a report on defining terrorism in a changing world.
- The Senate inquiry into the Fair Work Amendment (Right to Work from Home) Bill 2025 has issued its final report. In brief, both the government and the Coalition have rejected a universal right to request flexible working arrangements.
LATEST PODCAST: You may well enjoy The 9pm Cruel Philosophy of AI with Patrick Stokes, a philosopher, reprobate, and all-round cleverthinker. We ask whether AI is intelligent, or sentient, or has moral sensibilities. We discuss some fictional AI. We even talk about a bird — that’s nothing to do with AI though. It’s a fun conversation. Look for The 9pm Edict in your podcast app. Oh, and appearing very soon will be that chat with John Birmingham.
Elsewhere
- Anthropic, maker of the Claude series of AI models, has just completed history’s largest IPO, and once the dust settled the company was deemed to be valued at US$1.76 trillion after briefly hitting US$@ trillion. That hasn’t halted a broader market slide. The ever-cynical Pivot to AI refers to it as Anthropic’s “wizard plan” to loot your pension. Meanwhile Anthropic researcher Tao Lin says AI will make all humans permanently unemployable, so there’s that.
- The Quad nations — Australia, India, Japan, and the US — have agreed to advance an AI accelerator aimed at expanding access to trusted AI systems across the Indo-Pacific.
- Google leaned into AI and now it’s hurting its core business of selling ads.
- One of the funniest things I’ve seen recently has been Palantir’s response to the Democrats winning the US midterm elections. CEO Alex Carp says Palantir is “essentially progressive” and “I’ve been a Democrat my whole life... I know more about being a progressive than almost anybody on the planet.” Which is odd given that in April they posted what has been widely described as a technofascist manifesto.
- This Philadelphia woman stands in front of food delivery robots because its her “first amendment right to protest”.
Inquiries of note
- One for the deep nerds. Treasury has opened a consultation on the Consumer Data Right NPP service values alignment, which is probably as exciting as it sounds. NPP is the New Payments Platform, the system which allows you to transfer money instantly. Submissions close 16 December.
What’s next?
The House of Representatives returns this coming Monday 23 November to join the Senate in what’s scheduled to be the last week of parliamentary sittings for 2026. This may change, of course. There’s a lot for them to get through before Christmas.
The House of Reps draft legislation program currently shows nothing new of significance for this newsletter, although that may change too.
The Senate’s draft legislation program indicates that it’s expected to pass the Online Safety Amendment (Digital Duty of Care) Bill 2026, something the government really, really wants to happen.
DOES SOMETHING IN THE EMAIL LOOK WRONG? Let me know. If there’s ever a factual error, editing mistake, or confusing typo, it’ll be corrected in the web archives.