Katherine Mansted (@KMansted) header
Katherine Mansted (@KMansted) profile picture

Tech, strategy, info, democracy | @CyberCX @NSC_ANU | via: @SecureDemocracy @BelferCenter @MonashScholars

Joined December 2014
Photo by Katherine Mansted (@KMansted)Photo by Katherine Mansted (@KMansted)Photo by Katherine Mansted (@KMansted)Photo by Katherine Mansted (@KMansted)Photo by Katherine Mansted (@KMansted)Photo by Katherine Mansted (@KMansted)Photo by Katherine Mansted (@KMansted)Photo by Katherine Mansted (@KMansted)Photo by Katherine Mansted (@KMansted)Photo by Katherine Mansted (@KMansted)
A real privilege to be part of @ABCTV #4corners’ meticulously researched piece on cyber criminals hurting Australians & the cyber dimension of Russia’s war in 🇺🇦
CyberCX partnered with ABC’s @4Corners for their latest investigation, Hackers Inc, exposing the operations of the #cyber gangs targeting Australia. Watch the full #4corners episode to hear more expert insights from @KMansted: iview.abc.net.au/video/NC230…
3
8
1
38
7,520
Katherine Mansted retweeted
Oh, come on. Look at least use this as a sign, folks. This stuff isn’t even credible even as a bad movie plot. Yes actual risks are being ignored while we are misled because of dearth of technical experts on the journalism and politics side and because people believe all this.
OpenAI's Noam Brown says air-gapping the computers may not stop a misaligned AI, because two air-gapped machines can still talk by running a CPU hot and reading the temperature change "But I think the major takeaway from the incident is that people underestimated the AI. And we never want to be in a situation again where we underestimate the AI. It's a weird world, because AI progress is so fast that people are consistently underestimating the AI." "So to be in a situation where you don't underestimate it again, when it comes to safety and alignment, you have to have a very, very, very high bar." "You could even go as far as to say, "Well, we should air gap the computers." And I'm not convinced that that would be sufficient." "There are studies, and this is mostly academic, where you can have two computers next to each other that are air-gapped and they're still able to communicate with each other because they have temperature sensors." "One of them is able to run their CPU really hot, and then the other one can actually detect the temperature change, and then that actually gives them a mechanism to communicate." _________ Link and more key quotes from OpenAI's safety related conversations: firesidealpha.substack.com/p…
39
57
4
400
61,919
Katherine Mansted retweeted
You have noticed it. ChatGPT feels dumber than it used to. Your prompts that worked six months ago produce worse results now. The writing sounds flatter. The ideas sound safer. The internet itself feels like it is shrinking. Every article reads the same. Every email sounds the same. Every answer sounds like it was written by the same voice. You thought it was you. It is not you. Researchers at Oxford and Cambridge published a paper in Nature proving what is happening. They call it Model Collapse. Here is the mechanism in one sentence. AI trained on AI-generated data gets dumber every generation until it forgets what real human data looked like. The internet is filling with AI-generated content. Blog posts. Articles. Reviews. Comments. Social media. AI companies scrape the internet to train the next generation of models. Which means the next generation of AI is being trained on the output of the current generation. Each cycle loses information. Not randomly. It loses the rarest, most unusual, most creative parts first. The researchers call these the "tails of the distribution." The weird ideas. The unexpected perspectives. The things that made the internet feel human. Those disappear first. What remains is the average. The safe. The expected. The bland. Then the next generation trains on that. And loses more. And the next generation trains on that. And loses more. The researchers proved this is not a slow decline. Major degradation happens within just a few iterations. Even when some of the original human data is preserved. They tested it on large language models. On image generators. On statistical models. The pattern was the same every time. The output converges toward a narrow, flattened version of reality that looks nothing like the original data. The lead researcher put it plainly. "Large language models are like fire. A useful tool. But one that pollutes the environment." The pollution is invisible. You cannot see which sentence on the internet was written by a human and which was written by AI. Neither can the AI that is about to train on it. And once the tails are gone, they do not come back. The damage is irreversible. This is not a prediction anymore. It is a diagnosis. The internet you grew up on was built by humans writing things no algorithm would have written. Strange, personal, imperfect, alive. That internet is being diluted. One generation of AI at a time. And the models trained on what remains are learning a smaller and smaller version of the world. Model Collapse is not a technical problem. It is a cultural one. The thing that made the internet worth reading is the thing that disappears first.
1,113
6,235
1,370
17,447
2,313,138
The surge in AI-written speeches in Britain's House of Commons visualised:
Tom Tugendhat calls out the rise of MPs using ChatGPT to write speeches 👏🏻👏🏻 “This place has become absurd”
81
1,229
176
10,769
1,302,284
Experts on a @SkyNewsAust special event 'The War Cabinet', including CyberCX's @KMansted, have warned the risk of war is more significant than the public is being told, with Australia unable to sit out a regional conflict. ▶️ Watch the full discussion here: skynews.com.au/australia-new…
3
3
431
CyberCX's @KMansted will join an exclusive one-hour special event from @SkyNewsAust - The War Cabinet - a deep dive into Australia’s defence capabilities and our region’s evolving security dynamics.
1
2
5
425
NSC media round-up 📢 🌎@Dilnoza_U on how a Trump led US is shifting international order tinyurl.com/54v7ab7d 👥Ian Kemish reflects on what the international community can learn from the Srebrenica massacre in Bosnia, 30 years on tinyurl.com/4fd4uxvb 🤖@KMansted recommends exercising caution when co-operating with China on AI tinyurl.com/25495y27 🇦🇺@Rory_Medcalf provides early insights from NSC Community Consultations and explains the many forms of threats that confront our nation tinyurl.com/4x23an9r Read more from NSC experts ➡️ nsc.anu.edu.au/publications-…
3
1
12
1,312
As an Israel-Iran ceasefire hangs in the balance, CyberCX's @KMansted told @SkyNewsAust while military action is capturing the airwaves, there has also been a persistent and ongoing cyber dimension since the beginning of the conflict.
1
7
521
Katherine Mansted retweeted
Why the Australian Government should release its own intelligence review - before the election is called. Full unpaywalled text of my recent opinion piece in The Sydney Morning Herald and The Age. And yes, there is a concerning Trump angle nsc.anu.edu.au/content-centr…
3
25
1
44
4,515
Joining The Conversation Hour on ABC Radio Melbourne this morning, CyberCX's @KMansted said Australia might be behind the eight ball in realising just how vulnerable connected cars are.
2
4
487
Alastair @MacGibbon said CyberCX's discovery of the Green Cicada Network is one of the largest ever documented networks of inauthentic accounts found on a social media platform. news.com.au/technology/onlin…
1
11
12
2,638
This week CyberCX Intelligence identified a network of at least 5,000 inauthentic X accounts almost certainly controlled in concert by an artificial intelligence (AI) Large Language Model (LLM) based system. Download the full report here 👇🏽 cybercx.com.au/blog/cybercx-…
2
5
1
6
688
Katherine Mansted retweeted
Aussie cybersecurity company CyberCX say they’ve discovered a network of overseas bots designed to get us angry and divided, warning that the software is learning and growing in sophistication.  Executive Director of Cyber Intelligence at CyberCX Katherine Manstead explains.
6
3
10
3,964
Speaking to ABC @RadioNational this morning, CyberCX's @KMansted outlined how our Cyber Intelligence team uncovered one of the largest ever networks of AI-controlled fake social media accounts. You can listen to the full interview here: abc.net.au/listen/programs/r…
2
4
721
Katherine Mansted retweeted
“In Canberra journalism circles there is a saying that if you want to know what the Australians are doing on national security, ask the Americans. It’s a reflection of the Australian public service’s culture of secrecy relative to the US” capitalbrief.com/article/aus…
15
50
6
123
16,697
Speaking on ABC @RadioNational's program, The Money, CyberCX's @KMansted said the biggest IT outage in history exposed some hard truths about cyberspace. Listen to the in depth podcast here: abc.net.au/listen/programs/t…
1
1
348
Senior official Hugh Jeffrey takes on AUKUS criticisms. Just a hint of getting things off his chest. Pleasingly plain language including on transparent oceans ("bunk!") sovereignty anxieties ("zombie assertions") + AUKUS agonising ("perpetual ambivalence") podcasts.apple.com/am/podcas…
3
14
42
12,662
As organisations around the globe work to recover from the Crowdstrike outage, CyberCX's @KMansted told @7NewsAustralia this has been a sharp wake up call about the fragility of our digital ecosystem. Watch the full news story here: youtube.com/watch?v=DkSCFzHg…
4
1
7
1,340
CyberCX recognises the tireless, ongoing efforts of IT and cyber teams across all sectors of the economy as organisations, large and small, work to restore services impacted by a global systems outage. Every organisation has been impacted differently and will have different pathways and timeframes to fully recover from this outage. We can therefore expect some disruption to continue into the days and weeks ahead. As organisations grapple with this challenge, we urge people to think carefully before making significant changes to their cyber security arrangements. We strongly encourage organisations not to disable or remove Crowdstrike products from their environment without viable alternative measures already in place. Any such changes organisations choose to make should be cautiously and thoroughly considered. CyberCX is also aware of increased threats from scammers and malicious actors seeking to exploit the ongoing confusion, including attempts to impersonate cyber security professionals promising help to overstretched technical teams. As the situation evolves, organisations should be vigilant against such threats, report suspicious activity, and work with established and trusted cyber security partners. CyberCX continues to support our customers and partners through this time and will update them directly as we learn more. Links to further information and advice: Australian Signals Directorate advisory: cyber.gov.au/about-us/view-a… Crowdstrike blog with technical details of outage: crowdstrike.com/blog/technic… Crowdstrike blog outlining potential malicious activity: crowdstrike.com/blog/falcon-… Scamwatch report form: scamwatch.gov.au/report-a-sc…
3
2
529