Priviso Live

Episode 83: South Africa shines

🎙️ Priviso Live is back — and this week's episode is packed.From a homegrown South African AI breakthrough to robots running a half-marathon in Beijing, this week's episode of Priviso Live covers the stories that matter most at the intersection of technology, security, and governance.Here's what Lyn, Stephen, and Kayla unpack this week:🇿🇦 MzansiLM: AI built for South Africa Researchers from Stellenbosch University and UCT have released MzansiLM, a publicly available language model trained on all eleven of South Africa's official languages. Nine of those are considered low-resource in the AI world, meaning the big global models have largely ignored them. This is a significant step toward AI that actually works for South African citizens in healthcare, education, and government services. It is freely available on HuggingFace.🔬 Wits makes global waves in quantum computing Professor Andrew Forbes and his team at the University of the Witwatersrand have published landmark research in Nature Photonics, engineering quantum states that remain stable even in noisy real-world conditions. The implications for cryptography, secure communications, and financial modelling are considerable, and it feeds directly into South Africa's national Quantum Roadmap.🛡️ NIST changes how it handles CVEs, and your vulnerability programme should take note CVE submissions have grown 263% since 2020. NIST can no longer enrich every vulnerability. Security teams need to shift to risk-based vulnerability management, now more than ever.🤖 Google's Gemini AI blocked 8.3 billion malicious ads in 2025 That is not a typo. AI versus AI, and the arms race is well and truly on.🍽️ Chef Robotics hits 100 million commercial servings Robotic food production is no longer a pilot project. It is mainstream.🏃 300 humanoid robots run a half-marathon in Beijing Yes, really. Twenty-one kilometres. Bipedal. 🎧 Catch the full episode on Spotify, Apple Podcasts, iHeartRadio, Samsung Podcasts, and YouTube.#PrivisoLive #InfoSec #AI #Governance #CyberSecurity #RiskManagement #SouthAfrica #Technology

Priviso Live: Episode 82 SD 480p

🎙️ PRIVISO LIVE | Episode 82 is out now.Three stories this week that connect in ways you might not expect. All of them matter to every information security and technology professional in South Africa.⚛️ The Quantum Time Bomb is Already Ticking"Harvest Now, Decrypt Later" is not a future threat. Nation-state actors are collecting your encrypted data today, banking on quantum computing to make it readable by the 2030s. Only 3% of organisations globally have implemented all leading quantum-resistant measures. South Africa has no national quantum defence programme. The adversary is already collecting. Are your boardrooms paying attention?🦋 The AI Too Dangerous to ReleaseOn 7 April 2026, Anthropic announced Project Glasswing, a cybersecurity initiative built around an AI model so capable of finding zero-day vulnerabilities that they refused to release it publicly. Claude Mythos Preview found a 27-year-old flaw in OpenBSD, a 17-year-old remote code execution bug in FreeBSD, and thousands more, across every major operating system and browser. The fix is deliberate. The question for Africa is whether we have a seat close enough to the glass.💔 The Price We Are PayingA California jury found Meta and YouTube liable for engineering platforms to be addictive, awarding $6 million in a case that is the first of 2,000 pending. On Valentine's Day, 800,000 people grieved the retirement of a chatbot. Eight lawsuits allege AI companion responses contributed to suicides. McLuhan, Ellul, and Heidegger warned us. We did not listen.#PrivisoLive #InformationSecurity #QuantumComputing #AI #Cybersecurity #SouthAfrica #ICTGovernance #AIPolicy

Priviso Live Episode 81: Outlook in Outer Space

We open with something that should be on every CISO's radar: the Iran-US-Israel conflict has gone fully hybrid. Conventional warfare and cyberattacks are now inseparable. Investigators tracked nearly 5,800 cyberattacks from close to 50 Iran-aligned threat groups, targeting data centres, critical infrastructure, and yes, even a major medical technology company. If you think geopolitical cyber spillover doesn't affect you, think again.⚡ From there, we dive into RSAC 2026. Google Mandiant's M-Trends report confirmed that attack propagation time has dropped from eight hours to 22 seconds. A human-only security response is no longer mathematically possible. We also unpack the AI Tragedy of the Commons: why the rush to deploy agentic AI is creating attack surfaces that most organisations haven't even begun to map. Machine identities now outnumber human identities 80,000 to one.🤖 Then things get considerably more entertaining. Anthropic accidentally published 512,000 lines of Claude's source code to a public repository, and buried inside was an April Fool's Easter egg: a tamagotchi virtual pet called Claude Buddy, complete with 18 species, five rarity tiers, and a stat system tracking CHAOS and SNARK. The subsequent DMCA takedown of 8,100 GitHub repositories was also, apparently, an accident. You cannot make this up.📻 We also look at WRIT-FM: a live radio station run entirely by Claude AI. All programming. All DJ scripts. No humans required. The future of broadcasting, or a cautionary tale? We discuss.🚀 And we close with the most relatable story of the week: Artemis II astronauts, on their way to the Moon, calling Houston because Outlook wasn't working. Two instances. Neither functional. Rocket science, apparently, is no match for Microsoft.🔗 Watch Episode 81 now: [link in comments]#PrivisoLive #InformationSecurity #CyberWarfare #RSAC2026 #AgenticAI #Anthropic #Artemis #AI #Cybersecurity #SouthAfrica

Priviso Live Episode 80: Another Liberty breach

🎙️ Priviso Live is back, and in this week's episode we're covering four stories that matter right now to every information security, risk, and ICT professional in South Africa and beyond.📋 PAIA reports: The clock is tickingThe Information Regulator has issued its notice. The 2025/2026 PAIA Annual Report submission window opens on 1 April 2026 and closes 30 June. Public bodies, private companies, Information Officers, and Deputy IOs: this applies to you. And here's the catch: you cannot submit unless you are already registered on the eServices portal. Don't leave it to the last minute. Lyn and Stephen break down exactly what you need to do, and when.🔓 Liberty Group confirms a data breach againMarch 2026, and Liberty is back in the headlines for all the wrong reasons. An external threat actor extracted customer data, including ID numbers and names, and attempted extortion. Liberty refused to pay. The Information Regulator has requested an urgent meeting with Liberty's CEO. Kayla unpacks the POPIA Section 22 notification obligations, and Stephen asks the question every CISO should be able to answer: do you have an extortion response framework?🤖 OpenAI kills Sora, and walks away from DisneyLess than two years after launching to global fanfare, OpenAI has discontinued its Sora video generation platform and wound down its billion-dollar Disney partnership. The pivot? Robotics and agentic AI. We unpack what this means for organisations currently evaluating AI tools for marketing, training, and communications.😄 Opper AI Roundtable: 50 AI models, one question, zero conclusionsFree, fascinating, and faintly chaotic: Opper's AI Roundtable lets you put the same question to up to 50 models simultaneously and watch them debate. The site is found at https://opper.ai/ai-roundtable🎧 Find us on YouTube, Spotify, Apple Podcasts, and iHeartRadio.🌐 www.priviso.co.za#PrivisoLive #InformationSecurity #PAIA #POPIA #CyberSecurity #AI #DataBreach #Liberty #OpenAI #SouthAfrica #Compliance #ICT #RiskManagement #ArtificialIntelligence

Episode 79: Meta job losses

This week on Priviso Live, we dive into the stories shaping the future of technology, security, and digital rights in South Africa and ask the question every IT and security professional should be sitting with: as AI accelerates into every corner of our organisations, who is accountable when it goes wrong?Microsoft 365 E7 – The Frontier Suite | Microsoft has bundled Copilot, Agent 365, and the Entra security suite into one enterprise licence at $99/user/month. We break down what’s in the box, what it means for South African IT budgets, and why your Microsoft rep is about to become very persistent.Meta’s AI Pivot and Mass Layoffs | Up to 16,000 jobs at risk as Meta doubles down on a $135 billion AI spend. Markets cheered. Workers didn’t. We unpack what this signals for the broader tech sector — and for anyone in IT who isn’t yet upskilling in AI.South Africa’s Data Breach Crisis | A ransomware group exfiltrated 3.8TB of Gauteng citizen data: 3.6 million files, including ID documents and passports and listed them on the dark web for $25,000. Meanwhile, a new report confirms South Africa suffers a data breach every three hours, with 90% classified as preventable.Lessons from the Olympics and World Cup | How mega-events prepare for nation-state cyberattacks, and what your organisation can steal from their incident response playbook.The AI Incident Database | More than 1,000 documented cases of AI causing real-world harm. We introduce incidentdatabase.ai and explain why every practitioner deploying AI should have it bookmarked.Recorded on Human Rights Day — and the conversation about privacy, dignity, and accountability in the digital age has never felt more urgent.#PrivisoLive #InformationSecurity #AI #DataPrivacy #CyberSecurity #SouthAfrica #HumanRightsDay

Priviso Live Episode 78

🦞 A lobster just moved markets by $50 billion. No, really.This week on Priviso Live, we unpack the stories shaping the information security and AI landscape, and trust us, it's a packed episode.🇿🇦 South Africa's Draft AI Policy is heading to Cabinet. The 60-day public consultation gazette is expected this month, and the clock is ticking. Five core pillars. No single AI regulator. A sector-specific model that puts POPIA, FSCA, and King V front and centre. If you're in tech, compliance, or security, this one's not optional viewing.🌍 Africa is no longer playing it safe with AI governance. Forty-four countries now have data protection laws in force. Nigeria's AI Bill lands this month. The African Union's Continental AI Strategy is in Phase 1. The grace period is officially over.🔐 Cyber risk topped the Allianz Risk Barometer for the fifth year running, hitting its highest-ever score. And AI just made the biggest single jump in the barometer's 15-year history, surging from #10 to #2. Ransomware. Prompt injection. Shadow agents. The threat landscape is getting smarter, faster.🤖 AI agents are your organisation's new super-users. They access files. Approve transactions. Trigger workflows. At machine speed. Are you governing them like the privilege accounts they actually are?🦞 And then there's QClaw. Built on open-source OpenClaw. Integrated with WeChat . Led by a 26-year-old with five engineers and zero marketing budget. It went viral before it launched, and added $50 billion to Tencent market cap in a single day.Oh, and a 16-year-old just announced he's using OpenClaw to make $50,000 in six months. The future is not waiting.▶️ Priviso Live new episode is out now.🔔 Follow to stay sharp.#PrivisoLive #CyberSecurity #ArtificialIntelligence #AIGovernance #Privacy #SouthAfrica #AIPolicy #AI #InfoSec #AgenticAI #Africa #DataProtection #Africa #Priviso

Priviso Live Episode 77

🎙️ Priviso Live is back, and this week's episode hits close to home.We're unpacking four stories that every security, risk, and IT professional needs on their radar right now.🌍 Cyber Poverty is a Security Problem The Cyber Poverty Line Institute makes a compelling case: when communities lack digital access, skills, and resources, the entire cyber ecosystem pays the price. We explore what "cyber poverty" really means, and the Cyber Inclusion Pledge that's asking business leaders to do something about it.🚪 An OpenAI Exec Just Walked Out the Door Caitlin Kalinowski, OpenAI's head of hardware and robotics, resigned after the company struck a deal with the US Pentagon — and she didn't stay quiet about why. Her concerns weren't just ethical. They were about governance. About process. About who defines the red lines, and when.🎭 AI Theatre is Real, and It's Expensive $1.4M in Copilot licenses. 47 users. A board that approved it in 11 minutes. Peter Girnus's satirical post hit 23 million impressions because everyone recognised the pattern. We break down why enterprise AI fails — and what doing it properly actually looks like.🔐 The Attack Surface You're Not Watching AI ecosystem vulnerabilities grew 15,000% between 2018 and 2025. In 2025 alone: over 1,000 AI-related CVEs. Girnus also shares the work being done at TrendAI's Zero Day Initiative: real vulnerability research, not board deck theatre, and the questions every CISO should be asking right now.💬 Which of these stories hits hardest for your organisation right now? Drop it below.#InfoSec #CyberSecurity #ArtificialIntelligence #RiskManagement #AIGovernance #PrivisoLive #CyberPoverty #EnterpriseTech #SouthAfricahttps://www.cyberpovertyline.org/the-pledge

Episode 76: Hostilities Commence

🚨 Ransomware at Home. AI in the War Room. Missiles in the Middle East.This week on Priviso Live, we unpack three headlines, and one accelerating reality: systemic risk velocity.💻 Another South African ransomware attackThe Land and Agricultural Development Bank of South Africa confirmed it has been hit. Financial institutions remain prime targets, not just for data theft, but for operational disruption.ATMs. Online banking. Payment systems. Trust.Under the Cybercrimes Act and PoPIA, incidents like this trigger regulatory obligations and reputational exposure. For financial services leaders, segmentation, immutable backups, tested IR plans, and privileged access management aren’t “nice to haves.” They’re existential controls.If your backup is domain-joined, it’s not a backup: it’s a hostage.🤖 AI governance under real pressureAnthropic reportedly refused Pentagon demands to remove safeguards around mass surveillance and autonomous weapons. Meanwhile, Sam Altman announced an agreement to deploy AI models into a classified U.S. military network, with contractual safety principles embedded.Human-in-the-loop. No domestic mass surveillance. Technical safeguards. Field Deployment Engineers. Cloud-only containment.This isn’t theoretical AI ethics anymore. This is enforceable governance under geopolitical strain.For practitioners, the lesson is clear:When AI enters high-consequence environments, safeguards must move from aspirational to operational. Logged. Auditable. Tested.🌍 Geopolitical escalation and cyber spilloverAs anticipated, joint American-Israeli operations against Iranian targets reportedly commence. Diplomatic negotiations collapse. Embassy staff authorised to depart.Historically, kinetic escalation correlates with cyber escalation. Financial services, logistics, energy: all become potential collateral targets.Even if the conflict isn’t local, the digital ripple effects are.🎯 The synthesis?Ransomware locally. AI acceleration globally. Geopolitical volatility internationally.Three headlines. One theme:Risk is moving faster.Static annual reviews won’t cut it.Continuous monitoring. Threat intelligence integration. Tabletop exercises. Crisis communications discipline.Because cyber risk doesn’t exist in isolation. It mirrors the real world.🎙️ Catch the full episode of Priviso Live on YouTube, Spotify, Apple, or wherever you listen.#CyberSecurity #AI #Governance #RiskManagement #PoPIA #Ransomware #Geopolitics #CISO