AI, security, and the next operating model.
Unsupervised Learning turns AI and cybersecurity news into practical mental models. It moves between solo arguments, product conversations, and security notes with a bias toward what changes next.
PRESS PLAY
Unsupervised Learning is a technology strategy show with a sense of urgency. It watches AI, cybersecurity, national security, productivity software, and culture as parts of the same transition rather than separate beats. The core mode is argument. A typical episode starts with a claim, tests it against workflows and incentives, then keeps reframing it until the practical consequence becomes hard to miss. The show is especially interested in AI as infrastructure: assistants, agents, harnesses, browser-native systems, and tools that turn expert judgment into repeatable work. It does not treat AI as just another software category. It treats it as a change in the interface between people and the world. That comes through in episodes about a single digital assistant, AI replacing parts of knowledge work, and judging systems by output instead of mechanism. The security coverage is similarly concrete. Recent roundups move from NPM issues and Chrome fixes to identity, browser attacks, poisoned sites for agents, breach pressure on CISOs, and software supply chain transparency. Interviews include conversations about Maze and Gamma, with questions that stay close to product architecture, workflow pain, and what current tools still miss. The style is fast and conversational. It can ramble, but the rambling usually has a job: getting from abstract trend to usable mental model. There are also personal edges, including health updates, format experiments, and reflections on what the audience finds valuable. That looseness makes the show feel less like a scripted briefing and more like an experienced technical thinker narrating the map as it changes. It is best for listeners who want prediction, disagreement, and practical framing in the same feed. It is not neutral wallpaper. It has a point of view, and the point of view is that preparation matters.
Made for: For security practitioners, technical founders, AI builders, and strategy-minded listeners who want fast analysis of where tools, threats, and work are moving.
What sets it apart: The show connects big AI claims to concrete workflows: browsers, agents, vulnerability triage, presentations, assistants, and security programs. Its edge is the repeated move from headline to practical takeaway.
Unsupervised Learning is about ideas and trends in Cybersecurity, National Security, AI, Technology, and Culture—and how best to upgrade ourselves to be ready for what's coming.
Based on 6 episodes we listened to · September 2026
Unsupervised Learning makes AI's consequences legible by connecting ambitious forecasts to system design, security failures, and practical choices.
Unsupervised Learning turns fast-moving technology arguments into concrete models for work, security, and daily life. Daniel Miesler moves between tightly argued solo essays, news-heavy standard editions, and interviews with builders such as Maze co-founder and CEO Harry…
Daniel Miesler won't let your 'AI can't do my job' argument pass without a cross-examination.
Press play if you want
Short reviews from the PlayNext desk, based on the episodes we processed.
A tight case for why personal AI may settle into one trusted front door.
This episode argues for a single digital assistant as the main entry point for agents, harnesses, and the context layer around them. Its point is simple enough to stick: people should not have to keep relearning interfaces when one trusted layer can speak, gesture, and route the…
Maze's approach is to break vulnerability work into smaller questions before asking AI to do the analysis.
Unsupervised Learning, with Harry Wetherald of Maze, lands best when it gets practical about system design. The episode makes a clear case for breaking problems into smaller questions, leaning on deterministic code wherever possible, and treating agents less like all-purpose…
A clear look at Gamma's effort to rethink presentation-making through an AI-assisted creation workflow.
Grant Lee turns a familiar workplace irritation into a bigger question: why do presentations still depend so much on boxes, alignment, fixed formats, and habits that have barely moved in decades?
A lab-results opening gives way to browser security, AI agents, and supply-chain risk.
Unsupervised Learning episode 497 opens with lab results, kidney concerns, creatine, and a long sedentary stretch during CHI and AI work, then shifts into the browser as the main security focus.
A personal AI build log that turns into a clear argument about AI, security, and organizational control.
This episode follows a personal AI build that has clearly been a source of delay and frustration for the host, then broadens into a case for how AI fits inside organizations, with people setting direction and systems taking on more of the planning and execution.
A brisk, topic-led look at judging AI reasoning through its work, with security and human judgment close behind.
OpenAI's o1-preview model gives this Technology episode a clean organizing question: how should AI reasoning be judged? The host builds from a paper where human experts, reviewing sources blind, preferred AI-generated research ideas, then makes the case that comparative human…
A blunt warning that AI pressure is already reaching knowledge work.
This episode argues that knowledge workers should not assume AI pressure stops at the edges of their jobs. It pushes back on the familiar idea that models are only text machines, or that human expertise will stay neatly protected, and instead asks what work actually demands from…
Deeply researched narrative dives into hacking history and cyber-law cases, hosted by Ran Levi.
Reported narratives on hackers, spyware, and data crimes from Dina Temple-Raston.
Dave, Joe, and Maria break down phishing kits, deepfakes, and social-engineering scams weekly.
IAPP's Jedidiah Bracy interviews the regulators and lawyers writing privacy and AI law.
A daily cybersecurity briefing hosted by Dave Bittner, pairing breaking incidents with a guest interview.
Two cybersecurity professionals discuss their key takeaways from RSA 2025, highlighting that the most valuable innovation and conversations occurred at smaller, specialized satellite events rather than the main conference floor.
Daniel Miessler hosts Unsupervised Learning episode 450, analyzing OpenAI's o1-preview model and the path to AGI.
The host of Unsupervised Learning argues forcefully against the common narrative that AI cannot replace human knowledge workers.
The host discusses shifting the podcast format to shorter, more focused segments on cybersecurity, AI, and technology trends to improve engagement and reduce exhaustion.
The host argues for a single digital assistant.
This episode challenges the popular narrative that AI won't replace knowledge workers, arguing this belief is both incorrect and dangerous because it prevents proper preparation for technological disruption.
This episode explores the concept of 'the great transition' in cybersecurity, AI, technology, and society, focusing on how knowledge is shifting from private to public due to advancements like LLMs.
The host discusses experimenting with podcast formats and audience preferences, particularly focusing on free-form tangents over structured news reading.
This episode argues that AI should be judged by its creative output, using a blues version of Eminem's 'Without Me' from the 1950s as a compelling example of AI-generated music.
The host talks with Harry Wetherald, co-founder at Maze, about an AI system for vulnerability management.
Grant Lee describes Gamma as "the anti-PowerPoint," a product he started a little over four years ago to rethink slides from the ground up.
Episode 497 opens on lab results, and the kidney number gets the most attention.
Episode 496 opens with the host talking about a new video on building a personal AI system using cloud code, and he sounds genuinely relieved that it is finally out after a long, messy process.
Daniel interviews Patrick Duffy from Material Security about cybersecurity trends, focusing on preventative controls, threat detection, and current attack methods like phishing.
This episode discusses cybersecurity trends, AI tools like Klein and Windsurf, and the host's personal AI infrastructure setup.
This episode of Unsupervised Learning discusses trends in cybersecurity, AI, technology, and society, while sharing personal reading updates and recommendations.