As GenAI tools and the LLMs behind them impact the daily lives of billions, this report examines whether these technologies can be trusted to keep users safe.
What you’ll learn:
- How LLMs respond to risky prompts from bad actors and vulnerable users
- Where current models show safety strengths and weaknesses
- Actionable steps to improve LLM safety and reduce harmful outcomes
Overview
In this first independent benchmarking report on the LLM safety landscape, ActiveFence’s subject-matter experts put leading models to the test. More than 20,000 prompts were used to analyze how six LLMs respond across seven major languages and four high-risk abuse areas: child exploitation, hate speech, self-harm, and misinformation. The report provides comparative insight into each model’s relative safety strengths and weaknesses, helping teams understand where gaps exist and where additional resources may be required.
What’s new from Alice
Cyber RL Benchmark v1.0: Measuring Frontier Cybersecurity Capability
Alice's Cyber RL Benchmark v1.0 tested 10 frontier models on 12 real cybersecurity tasks, with fresh code, real vulnerabilities, planted decoys, and objective scoring. The finding: no model can reliably cover the range of a security analyst's job.
LIVE from Black Hat Las Vegas: AI, Nation-States, and the Battlefield That Keeps Changing
Recorded live at Black Hat, Caroline Wong and Allie Mellen unpack why AI made nation-state attacks cheaper, why safety guardrails can work against you, and what the OpenAI-Hugging Face incident really exposed.
Virtual Fireside Chat: The TAKE IT DOWN Act, Six Months In - What's Changed on Deepfakes and NCII
Six months after the TAKE IT DOWN Act took effect, the NCII landscape looks different and even more complicated. Join Alice for a live fireside chat with Google's Nidhi Lahoti on what's actually changed, what hasn't, and where T&S fit into it all - Join us live on October 9th, 2pm EST.
5 Ways Your Third-Party CX Agent Gets Broken
Third-party CX agents create hidden liability. Learn the 5 attack patterns vendors miss and how WonderSuite closes the gap.
