machine learning
Auditing the Ask Astro LLM Q&A app
Today, we present the second of our open-source AI security audits: a look at security issues we found in an open-source retrieval augmented generation (RAG) application that could lead to chatbot output ...
Why SaaS Identity Abuse is This Year’s Ransomware
Let’s explore some of the details behind this escalating threat to SaaS applications, what may be driving it, and what you can do to better protect your SaaS footprint from these types ...
The Impending Identity Crisis Of Machines: Why We Need To Secure All Non-Human Identities, From Genai To Microservices And IOT
The digital landscape is no longer solely populated by human actors. Lurking beneath the surface is a silent legion – non-human or machine identities . These non-human identities encompass computers, mobile devices, ...
Understanding Apple’s On-Device and Server Foundation Models release
By Artem Dinaburg Earlier this week, at Apple’s WWDC, we finally witnessed Apple’s AI strategy. The videos and live demos were accompanied by two long-form releases: Apple’s Private Cloud Compute and Apple’s ...
PCC: Bold step forward, not without flaws
By Adelin Travers Earlier this week, Apple announced Private Cloud Compute (or PCC for short). Without deep context on the state of the art of Artificial Intelligence (AI) and Machine Learning (ML) ...
Recall ‘Delayed Indefinitely’ — Microsoft Privacy Disaster is Cut from Copilot+ PCs
Copilot Plus? More like Copilot Minus: Redmond realizes Recall requires radical rethink ...
Snowflake and the Continuing Identity Threat Detection Gap Across SaaS and Cloud
By Adam Koblentz In recent weeks, a new wide-scale identity security incident has been unfolding that is refocusing the spotlight on important questions such as: Why are account takeover, credential misuse, and ...
Exploiting ML models with pickle file attacks: Part 2
By Boyan Milanov In part 1, we introduced Sleepy Pickle, an attack that uses malicious pickle files to stealthily compromise ML models and carry out sophisticated attacks against end users. Here we ...
Exploiting ML models with pickle file attacks: Part 1
By Boyan Milanov We’ve developed a new hybrid machine learning (ML) model exploitation technique called Sleepy Pickle that takes advantage of the pervasive and notoriously insecure Pickle file format used to package ...
Announcing AI/ML safety and security trainings
By Michael D. Brown We are offering AI/ML safety and security training in summer and fall of this year! Recent advances in AI/ML technologies opened up a new world of possibilities for ...
