Skip to Content
Get started
Open search
Focus Scroll Background

Llm

RSS

Sophos Insights

LLM

AI

Exploit

vulnerability

Active Adversary

Pacific Rim

The vulnerability flood is here. Here’s what it means – and how to prepare

We can't control the pace of AI-driven vulnerability discovery, but we can control how fast we respond.
Ross McKerchar
A silver-white robot hand holding a silver tray on its fingertips, against a blurred background of a large room

The OpenClaw experiment is a warning shot for enterprise AI security

Threat Research

A computer hard disk on fire against a white background

Laughter in the dark: Tales of absurdity from the cyber frontline and what they taught us

Threat Research

A photograph of a hand reaching out to lock a prison cell with a bunch of keys

Locking it down: A new technique to prevent LLM jailbreaks

AI Research

A salt shaker lying flat on a blue tabletop, with the lid off and salt spilling out on to the table

Getting salty with LLMs: SophosAI unveils new defense against jailbreaking at CAMLIS 2025

AI Research

A hand holding a magnifying glass above several blocks containing pictures of triangles. The block directly under the magnifying glass is taller and has a larger triangle on it

Sophos AI at Black Hat USA ’25: Anomaly detection betrayed us, so we gave it a new job

Threat Research

A large toy robot holding the hand of a smaller one, on a white floor against a red-orange background

Small world: The revitalization of small AI models for cybersecurity

Threat Research

A row of white cubes on a blue background. The cubes contain checkmarks, with a yellow cube in the middle containing an alert symbol.

SophosAI at Black Hat USA ’25: Anomaly detection betrayed us, so we gave it a new job

Threat Research

shutterstock_asset-generation-6dc4f763-f8a7-4dff-a56b-92736d8c8d6c-1_edited

DeepSpeed: a tuning tool for large language models

AI Research

Flat,Line,Design,Website,Banner,Of,Learning,Process,,Brain,Process,

Sophos AI to present on how to defang malicious AI models at Black Hat Europe

AI Research