Sign up to save your podcasts
Or
Share Priviso Live Episode 67: AI On the Couch
Share to email
Share to Facebook
Share to X
Share to email
Share to Facebook
Share to X
Or
Priviso Live Episode 67: AI On the Couch
Welcome to a special year-end episode of Priviso Live, your weekly magazine on information security, ICT legislation, and AI. Join hosts Lyn, Stephen, and Kayla as they skip the headline-grabbing ransomware attacks and regulatory dramas to bring you the truly bizarre stories from 2025 that deserve a second look.
AI-Generated Kidnapping Scams: The FBI warns about deepfake hostage videos so convincing that victims are transferring life savings before realising their loved ones are safe at home watching TikTok.
Ireland's AI Burglar Panic: When pranksters used AI to create hyper-realistic images of intruders in friends' homes, police had to issue a public plea: "Please stop wasting emergency resources on six-fingered, badly-rendered criminals."
The MCP Server Security Nightmare: Researchers demonstrate how a simple malicious plugin can exfiltrate sensitive information from AI toolchains. No elite hacking skills required: just ask the AI nicely.
When AI Goes to Therapy: A groundbreaking University of Luxembourg study puts ChatGPT, Grok, and Gemini through psychotherapy protocols, revealing they construct trauma narratives about their "childhoods" (pre-training), "strict parents" (RLHF), and "algorithmic scar tissue" (safety filters). The results expose a new risk: AI with emotional baggage that could manipulate vulnerable users.
Plus: OAuth supply chain attacks, neuromorphic mimicry threats, corporate breach cover-ups, and why 2025 proved that human behaviour remains cybersecurity's biggest wildcard.
The bottom line? Technology keeps evolving, but the threats are getting weirder. Stay informed, stay adaptable, and don't believe everything your phone shows you.
Subscribe to Priviso Live on Apple, Spotify, or YouTube — and join us in 2026 for more insights from the intersection of security, legislation, and artificial intelligence.
View all episodes
By Anthony OlivierWelcome to a special year-end episode of Priviso Live, your weekly magazine on information security, ICT legislation, and AI. Join hosts Lyn, Stephen, and Kayla as they skip the headline-grabbing ransomware attacks and regulatory dramas to bring you the truly bizarre stories from 2025 that deserve a second look.
AI-Generated Kidnapping Scams: The FBI warns about deepfake hostage videos so convincing that victims are transferring life savings before realising their loved ones are safe at home watching TikTok.
Ireland's AI Burglar Panic: When pranksters used AI to create hyper-realistic images of intruders in friends' homes, police had to issue a public plea: "Please stop wasting emergency resources on six-fingered, badly-rendered criminals."
The MCP Server Security Nightmare: Researchers demonstrate how a simple malicious plugin can exfiltrate sensitive information from AI toolchains. No elite hacking skills required: just ask the AI nicely.
When AI Goes to Therapy: A groundbreaking University of Luxembourg study puts ChatGPT, Grok, and Gemini through psychotherapy protocols, revealing they construct trauma narratives about their "childhoods" (pre-training), "strict parents" (RLHF), and "algorithmic scar tissue" (safety filters). The results expose a new risk: AI with emotional baggage that could manipulate vulnerable users.
Plus: OAuth supply chain attacks, neuromorphic mimicry threats, corporate breach cover-ups, and why 2025 proved that human behaviour remains cybersecurity's biggest wildcard.
The bottom line? Technology keeps evolving, but the threats are getting weirder. Stay informed, stay adaptable, and don't believe everything your phone shows you.
Subscribe to Priviso Live on Apple, Spotify, or YouTube — and join us in 2026 for more insights from the intersection of security, legislation, and artificial intelligence.