Skip to content

AI broke the cybersecurity world. Now what?

Decoder

Hosted by Nilay Patel

1h 3m
Transcript

Original podcast

Where to listen

Summary

Written by a model from the transcript; not yet reviewed.

This episode of Decoder features a discussion between the host Nilay Patel and cybersecurity expert Alex Stamos about how artificial intelligence has upended cybersecurity and internet defense. The guest explains that the traditional multi-billion-dollar cybersecurity industry, built on a hierarchy of rare human hacking skills and slow attack cycles, is essentially over because open-weight models have automated the kill chain and lowered the cost of attacks to near zero. They argue that we are entering a digital dark age characterized by continuous breaches, automated vulnerability discovery, and a massive imbalance between attackers and defenders. To survive this era, the tech industry must move past passive bug patching and embrace active red teaming, cost-efficient operational defense, and the complete rewriting of legacy infrastructure like the Linux kernel into memory-safe languages. Finally, the conversation touches on the broader societal impacts of AI, exploring why verifying real content cryptographically is easier than detecting fake media and how technology companies must rethink their approach to trust and safety.

  • Organizations must actively red team themselves using AI tools to find vulnerabilities before malicious actors exploit them. 33:52
  • The traditional software security model of patching bugs is unsustainable, and foundational codebases like the Linux kernel should be rewritten in memory-safe languages. 25:46
  • Verifying real content cryptographically is more effective and attainable than attempting to detect fake media through probabilistic methods. 57:43

Chapters

  1. 0:00
    Show Sponsorship and Host Introduction

    The episode opens with Engine’s sponsorship announcement followed by host Neil Apatel introducing the show and guest Alex Stamos.

  2. 0:47
    Setting the Stage: AI and Cybersecurity Crisis

    Neil frames the discussion around AI’s disruptive impact on the cybersecurity industry and previews the episode’s focus on the current mess.

  3. 5:42
    AI-Enabled Kill Chain and Automated Exploits

    The conversation shifts to how AI inverts the traditional kill chain, automating vulnerability discovery and exploit creation in real time.

  4. 6:14
    Open-Source AI Tools Fueling Attacks

    Alex discusses the rise of open-source AI tools like RTex that enable widespread, low-cost attacks on organizations worldwide.

  5. 8:00
    Rewriting Legacy Code: The Linux Kernel Proposal

    Alex proposes a bold rewrite of the Linux kernel in a safer language to reduce long-term vulnerability exposure.

  6. 14:00
    Future Outlook: Slowing Down or Accelerating AI Security

    Neil and Alex debate whether the industry should slow down to catch up with defensive capabilities or continue accelerating AI development.

  7. 16:32
    Economic Shifts in Cyber Defense

    The hosts examine how AI changes the cost dynamics between attackers and defenders, emphasizing the need for cheaper, automated defense solutions.

  8. 33:57
    Operational Security Automation and Red Teaming

    Discussion turns to automating SOC work, AI-driven red teaming, and the challenges of adopting these tools in real organizations.

  9. 43:56
    Closing Thoughts and Takeaways

    The episode concludes with final reflections on the dark age of cybersecurity, the need for proactive defense, and a call to action for the industry.

  10. 48:04
    AI Risk Focus Shifts from Apocalypse to Human Uplift

    The conversation moves from P‑doom concerns to the broader human‑uplift risks of AI misuse.

  11. 50:11
    Balancing AI Control with Human Benefit

    Discussion on ensuring AI systems serve humans without adopting human‑like personalities.

  12. 51:22
    Facebook’s Content Moderation Challenges

    The host brings up Meta’s struggles with misinformation, deepfakes, and copyright issues on Instagram.

Highlights

  • 2:50

    The End of Traditional Cybersecurityinsight

    The guest explains that the security industry as known for decades is over because AI-boosted waves of automated hacks, breaches, and continuous attacks have inverted the cyber kill chain.

  • 3:22

    Chained Hacks and Continuous Breachesstory

    The guest details a real-world supply chain attack where stolen API keys from an AI provider allowed attackers to breach Cursor and push a bad plugin to thousands of companies.

  • 4:34

    Automating the Cyber Kill Chaininsight

    Adversaries can now run AI systems that find bugs and write exploits in real time, bypassing the need for human vulnerability researchers and large operational teams.

  • 11:04

    The Collapse of Security Economicsinsight

    The historical cybersecurity economy relied on a small group of elite experts whose skills could be hired for defense, but AI has diffused those capabilities to script kiddies.

  • 26:14

    Rewriting the Linux Kernelstrategy

    The guest argues that major tech companies should spend millions to fork and rewrite the aging Linux kernel in a memory-safe language rather than endlessly patching legacy code.

  • 33:52

    Mandatory AI Red Teamingstrategy

    Companies must immediately begin red teaming their systems with AI tools, accepting legal and operational risks to find vulnerabilities before young hackers do.

  • 57:43

    Verifying Real Media Cryptographicallyinsight

    Detecting fake AI-generated content is probabilistic and approaches impossibility, whereas verifying real content through camera sensor data and device chains of custody can be cryptographically perfect.

Quotes

  • “The multibillion-dollar cybersecurity industry, as we've known it, is essentially over due to the unrelenting AI-boosted waves of hacks, breaches and attacks.”
    Nilay Patel · 1:07
  • “So the system that everybody's talking about is not the only one.”
    Unnamed speaker · 6:09
  • “The cost per token for attackers is going towards zero. And the cost for defenders is staying quite high.”
    Unnamed speaker · 16:32
  • “I think they have teams who have been this entire time, working very hard on the big-picture cyber risk.”
    Unnamed speaker · 48:04
  • “I think detecting what's fake is always going to be probabilistic, and we're going to asymptotically approach impossibility, whereas detecting what's real can be cryptographically perfect.”
    Unnamed speaker · 57:40
  • “Text is much, text is much harder because you just have so much less data, right?”
    Unnamed speaker · 57:10
  • “I think AI is going to enable individuals, right? Like, we're gonna see a huge explosion of new companies.”
    Unnamed speaker · 59:36
  • Engine 0:00–0:42

About this episode

I’ve always enjoyed talking to security expert Alex Stamos because he’s so direct — and he doesn't mince words about how the multibillion dollar cybersecurity industry as we’ve known it is all but doomed thanks to the unrelenting, AI-boosted waves of hacks, breaches, and attacks. That much disruption means a lot of places where bad guys can get in while everyone else is left scrambling. Alex isn’t a doomer in the long run, but he does think we've got a dark age to get through before things get better. Links: Attack of the killer script kiddies | The Verge Alex Stamos on AI's coming cybersecurity dark age | Ground Level AI Palo Alto’s CEO Said ‘AI Must Fight AI’ | Palo Alto Networks Cybersecurity vets protest ‘dangerous’ ban on Anthropic’s Mythos | TechCrunch Binance: $6.3B heist used fake job interviews, internal AI tool | The Verge A thousand fewer paths to root | OpenAI Subscribe to The Verge to access the ad-free version of Decoder! Credits: Decoder is a production of The Verge and part of the Vox Media Podcast Network. Decoder is produced by Greg Ott, Kate Cox, and Nick Statt. Our editors are Xander Adams and Ursa Wright. Our editorial director is Kevin McShane. The Decoder music is by Breakmaster Cylinder. Learn more about your ad choices. Visit podcastchoices.com/adchoices

Transcript

0:00
1:03:24

11,885 words · 3 speakers

Speaker names marked are inferred from speaking time and the show’s known hosts — they are not voice-verified.

Topics

  • Linux kernel
  • Instagram
  • OpenAI
  • Anthropic
  • Facebook
  • Microsoft
  • Meta
  • iPhone
  • GlibC
  • Patch Tuesday

Page updated · transcript and summary are machine-generated