Kevin Eykholt

Taesung Lee

Ambrish Rawat

Killian Levacher

Mathieu Sinn

Abigail Goldsteen

Pin-Yu Chen

Kim Martineau

Research

Technical note

Deep Dive

Release

Explainer

Did an AI write that? If so, which one? Introducing the new field of AI forensics

An open-source toolkit for debugging AI models of all data types

Researchers develop defenses against deep learning hack attacks

AI goes anonymous during training to boost privacy protection

What is red teaming for generative AI?

Securing AI systems with adversarial robustness

Manipulating stock prices with an adversarial tweet

We're working on defending AI systems from attacks, creating new attacks to uncover better defenses, and a formal AI robustness certification.

AI Robustness

ICML 2024

COLM 2024

AIES 2024

NeurIPS 2024

ACML 2024

Trustworthy AI

Fairness, Accountability, Transparency

Adversarial Robustness and Privacy

Foundation Models

Machine Learning

Natural Language Processing

Generative AI

Security

Data and AI Security

Explainable AI

AI Testing

AI can help sift through the noise to identify significant patterns in the increasingly large pools of data generated by enterprises. We’re reimagining IT operations, using AI to automate core processes, better detection and diagnosis, preformance monitoring, and explainable action recommendations.

AI for IT

We’re designing tools to help ensure that AI systems are trustworthy, reliable and can optimize business processes. We create tests to simulate real-life scenarios and localize the faults in AI systems. We’re working on automating testing, debugging, and repairing AI models across a wide range of scenarios.

We’re designing tools to help ensure that AI systems are trustworthy, reliable and can optimize business processes. 

Even advanced AI systems can be vulnerable to adversarial attacks. We’re making tools to protect AI and certify its robustness, including quantifying the vulnerability of neural networks and designing new attacks to make better defenses. And we’re helping AI systems adhere to privacy requirements.

As organizations move to the hybrid cloud, they must protect sensitive data and comply with regulations that allow them to take advantage of AI. We’re designing systems to monitor and protect data, building trust in AI through robust evaluation, certification, and hardening against attacks.

Designing systems to monitor and protect sensitive data for robust and secure AI systems on the hybrid cloud.

We’re building the next wave of tools and methodologies to help security and operations teams detect, understand, and respond to advanced cybersecurity threats and attacks on their infrastructure and in the cloud, with automated threat detection, investigation, and deflection capabilities.

Threat Management

Building the next wave of tools to help security teams detect and deflect against advanced cybersecurity threats.

Diagramming adversarial robustness concepts

Adversarial Robustness and Privacy

Our work

What is red teaming for generative AI?

An open-source toolkit for debugging AI models of all data types

Did an AI write that? If so, which one? Introducing the new field of AI forensics

Manipulating stock prices with an adversarial tweet

Securing AI systems with adversarial robustness

Researchers develop defenses against deep learning hack attacks

Publications

Attack Atlas: A Practitioner's Perspective on Challenges and Pitfalls in Red Teaming GenAI

Privacy without Noisy Gradients: Slicing Mechanism for Generative Model Training

Membership Inference Attacks Against Time-Series Models

MoJE: Mixture of Jailbreak Experts, Naive Tabular Classifiers as Guard for Prompt Attacks

On Robustness-Accuracy Characterization of Language Models using Synthetic Datasets

Be Your Own Neighborhood: Detecting Adversarial Examples by the Neighborhood Relations Built on Self-Supervised Learning

Related topics