Artwork

Contenuto fornito da Ryan Staley. Tutti i contenuti dei podcast, inclusi episodi, grafica e descrizioni dei podcast, vengono caricati e forniti direttamente da Ryan Staley 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 !

Plugging the Security Holes in Your AI Pipeline with Chris "Tito" Sestito CEO and Co-Founder of Hidden Layer

38:41
 
Condividi
 

Manage episode 398417776 series 3402321
Contenuto fornito da Ryan Staley. Tutti i contenuti dei podcast, inclusi episodi, grafica e descrizioni dei podcast, vengono caricati e forniti direttamente da Ryan Staley 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.

Ryan interviews Chris "Tito" Tito, co-founder and CEO of Hidden Layer, a company providing security solutions for AI models. Tito outlines Hidden Layer's full platform to secure AI, including scanning models for malware, monitoring model behaviour to catch attacks, and proxying API calls to block abuse. Tito then discusses various impressive enterprise use cases he's seen and talks about the common mistake businesses are making.

Join 2,500+ readers getting weekly practical guidance to scale themselves and their companies using Artificial Intelligence and Revenue Cheat Codes.

Explore becoming Superhuman here: https://superhumanrevenue.beehiiv.com/

KEY TAKEAWAYS

  • Pre-trained open-source models often get reused internally but may contain backdoors if not scanned thoroughly first.
  • Current cybersecurity tools don't inspect model code so can't catch malware that is desperately needed for specialized AI security.
  • Tito stresses that while AI promises much value, it requires thoughtful implementation with appropriate safeguards.
  • Models deployed into production pipelines are vulnerable to surrogate attacks and theft.
  • Safeguards are needed for both internally hosted models and external API-based ones.
  • Diverse, specialized use cases are extracting tremendous value from AI across industries.
  • However, many teams lack the skills or processes to properly evaluate or fine-tune models.
  • Generative AI is increasingly being used for fraud, deception and reputation attacks.
  • The talent shortage, especially in adversarial ML, poses scaling challenges for AI security start-ups.

BEST MOMENTS

"We have found thousands of examples of models that have been loaded with malware where we've malware hiding in weights of models."

"Current cybersecurity stack is not even parsing models. It's like trying to point a windows product at a Mac file."

"If you have the inputs to the model and you have the outputs to the model, you basically have the training set and the labels."

"Every time we talk to a customer, we find a new [AI use case]. The world is getting really creative."

"There's just an enormous shortage of adversarial machine learning talent out there."

Ryan Staley

Founder and CEO

Whale Boss

ryan@whalesellingsystem.com

www.ryanstaley.io

Saas, Saas growth, Scale, Business Growth, B2b Saas, Saas Sales, Enterprise Saas, Business growth strategy, founder, ceo: https://www.whalesellingsystem.com/closingsecrets

  continue reading

405 episodi

Artwork
iconCondividi
 
Manage episode 398417776 series 3402321
Contenuto fornito da Ryan Staley. Tutti i contenuti dei podcast, inclusi episodi, grafica e descrizioni dei podcast, vengono caricati e forniti direttamente da Ryan Staley 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.

Ryan interviews Chris "Tito" Tito, co-founder and CEO of Hidden Layer, a company providing security solutions for AI models. Tito outlines Hidden Layer's full platform to secure AI, including scanning models for malware, monitoring model behaviour to catch attacks, and proxying API calls to block abuse. Tito then discusses various impressive enterprise use cases he's seen and talks about the common mistake businesses are making.

Join 2,500+ readers getting weekly practical guidance to scale themselves and their companies using Artificial Intelligence and Revenue Cheat Codes.

Explore becoming Superhuman here: https://superhumanrevenue.beehiiv.com/

KEY TAKEAWAYS

  • Pre-trained open-source models often get reused internally but may contain backdoors if not scanned thoroughly first.
  • Current cybersecurity tools don't inspect model code so can't catch malware that is desperately needed for specialized AI security.
  • Tito stresses that while AI promises much value, it requires thoughtful implementation with appropriate safeguards.
  • Models deployed into production pipelines are vulnerable to surrogate attacks and theft.
  • Safeguards are needed for both internally hosted models and external API-based ones.
  • Diverse, specialized use cases are extracting tremendous value from AI across industries.
  • However, many teams lack the skills or processes to properly evaluate or fine-tune models.
  • Generative AI is increasingly being used for fraud, deception and reputation attacks.
  • The talent shortage, especially in adversarial ML, poses scaling challenges for AI security start-ups.

BEST MOMENTS

"We have found thousands of examples of models that have been loaded with malware where we've malware hiding in weights of models."

"Current cybersecurity stack is not even parsing models. It's like trying to point a windows product at a Mac file."

"If you have the inputs to the model and you have the outputs to the model, you basically have the training set and the labels."

"Every time we talk to a customer, we find a new [AI use case]. The world is getting really creative."

"There's just an enormous shortage of adversarial machine learning talent out there."

Ryan Staley

Founder and CEO

Whale Boss

ryan@whalesellingsystem.com

www.ryanstaley.io

Saas, Saas growth, Scale, Business Growth, B2b Saas, Saas Sales, Enterprise Saas, Business growth strategy, founder, ceo: https://www.whalesellingsystem.com/closingsecrets

  continue reading

405 episodi

Tutti gli episodi

×
 
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