Artwork

Contenuto fornito da Hacker Valley Media. Tutti i contenuti dei podcast, inclusi episodi, grafica e descrizioni dei podcast, vengono caricati e forniti direttamente da Hacker Valley Media o dal partner della piattaforma podcast. Se ritieni che qualcuno stia utilizzando la tua opera protetta da copyright senza la tua autorizzazione, puoi seguire la procedura descritta qui https://it.player.fm/legal.
Player FM - App Podcast
Vai offline con l'app Player FM !

Building EDR for AI: Controlling Autonomous Agents Before They Go Rogue with Ron Eddings

19:36
 
Condividi
 

Manage episode 522307949 series 2675076
Contenuto fornito da Hacker Valley Media. Tutti i contenuti dei podcast, inclusi episodi, grafica e descrizioni dei podcast, vengono caricati e forniti direttamente da Hacker Valley Media o dal partner della piattaforma podcast. Se ritieni che qualcuno stia utilizzando la tua opera protetta da copyright senza la tua autorizzazione, puoi seguire la procedura descritta qui https://it.player.fm/legal.

AI agents aren't just reacting anymore, they're thinking, learning, and sometimes deleting your entire production database without asking. The real question isn't if your AI agent will be hacked, it's when, and whether you'll have the right hooks in place to stop it before it happens.

In this episode, Ron breaks down the ChatGPT Atlas vulnerability that shocked researchers, revealing how malicious prompts can turn AI assistants against their own users by bypassing safeguards and accessing file systems. He presents his new talk "Hooking Before Hacking," introducing a framework for applying EDR principles, prevention, detection, and response, to AI agents before they execute unauthorized commands. From pre-tool use hooks that catch malicious intent to one-time passwords that put humans back in the loop, this episode shares practical security controls you can implement today to prevent your AI agents from going rogue.

Impactful Moments:

00:00 - Introduction
02:00 - ChatGPT Atlas vulnerability exposed
04:00 - AI technology outpacing security guardrails
05:00 - Guardrail jailbreaks and prompt injection
06:00 - AI agents deleting production databases
07:00 - EDR principles for AI agents
09:00 - Pre-tool use hooks catch intention
11:00 - User prompt sanitization prevents leaks
14:00 - One-time passwords for agent workflows
16:00 - Automation mistakes across 10 years

Links:

Connect with Ron on LinkedIn: https://www.linkedin.com/in/ronaldeddings/

Check out the entire article here: https://www.yahoo.com/news/articles/cybersecurity-experts-warn-openai-chatgpt-101658986.html

GitHub Repository: https://hackervalley.com/hooking-before-hacking

See Ron's "Hooking Before Hacking" presentation slides here: http://hackervalley.com/hooking-before-hacking-presentation

Check out our website: https://hackervalley.com/

Upcoming events: https://www.hackervalley.com/livestreams

Love Hacker Valley Studio? Pick up some swag: https://store.hackervalley.com

Continue the conversation by joining our Discord: https://hackervalley.com/discord

Become a sponsor of the show to amplify your brand: https://hackervalley.com/work-with-us/

Join our creative mastermind and stand out as a cybersecurity professional: https://www.patreon.com/hackervalleystudio

  continue reading

405 episodi

Artwork
iconCondividi
 
Manage episode 522307949 series 2675076
Contenuto fornito da Hacker Valley Media. Tutti i contenuti dei podcast, inclusi episodi, grafica e descrizioni dei podcast, vengono caricati e forniti direttamente da Hacker Valley Media o dal partner della piattaforma podcast. Se ritieni che qualcuno stia utilizzando la tua opera protetta da copyright senza la tua autorizzazione, puoi seguire la procedura descritta qui https://it.player.fm/legal.

AI agents aren't just reacting anymore, they're thinking, learning, and sometimes deleting your entire production database without asking. The real question isn't if your AI agent will be hacked, it's when, and whether you'll have the right hooks in place to stop it before it happens.

In this episode, Ron breaks down the ChatGPT Atlas vulnerability that shocked researchers, revealing how malicious prompts can turn AI assistants against their own users by bypassing safeguards and accessing file systems. He presents his new talk "Hooking Before Hacking," introducing a framework for applying EDR principles, prevention, detection, and response, to AI agents before they execute unauthorized commands. From pre-tool use hooks that catch malicious intent to one-time passwords that put humans back in the loop, this episode shares practical security controls you can implement today to prevent your AI agents from going rogue.

Impactful Moments:

00:00 - Introduction
02:00 - ChatGPT Atlas vulnerability exposed
04:00 - AI technology outpacing security guardrails
05:00 - Guardrail jailbreaks and prompt injection
06:00 - AI agents deleting production databases
07:00 - EDR principles for AI agents
09:00 - Pre-tool use hooks catch intention
11:00 - User prompt sanitization prevents leaks
14:00 - One-time passwords for agent workflows
16:00 - Automation mistakes across 10 years

Links:

Connect with Ron on LinkedIn: https://www.linkedin.com/in/ronaldeddings/

Check out the entire article here: https://www.yahoo.com/news/articles/cybersecurity-experts-warn-openai-chatgpt-101658986.html

GitHub Repository: https://hackervalley.com/hooking-before-hacking

See Ron's "Hooking Before Hacking" presentation slides here: http://hackervalley.com/hooking-before-hacking-presentation

Check out our website: https://hackervalley.com/

Upcoming events: https://www.hackervalley.com/livestreams

Love Hacker Valley Studio? Pick up some swag: https://store.hackervalley.com

Continue the conversation by joining our Discord: https://hackervalley.com/discord

Become a sponsor of the show to amplify your brand: https://hackervalley.com/work-with-us/

Join our creative mastermind and stand out as a cybersecurity professional: https://www.patreon.com/hackervalleystudio

  continue reading

405 episodi

ทุกตอน

×
 
Loading …

Benvenuto su Player FM!

Player FM ricerca sul web podcast di alta qualità che tu possa goderti adesso. È la migliore app di podcast e funziona su Android, iPhone e web. Registrati per sincronizzare le iscrizioni su tutti i tuoi dispositivi.

 

Guida rapida

Ascolta questo spettacolo mentre esplori
Riproduci