Perspectives
Insights on AI security, governance, and real-world risk.
Learn how to prevent prompt injection, secure AI systems, and manage AI in production.
What AI Red Teaming Looks Like Outside the Lab
From Online Lie to Offline Riot: What Torre Pacheco Reveals About Coordinated Hate Campaigns
Violent anti-migrant riots in Spain were fueled by misinformation and hate networks operating across social media. Learn how disinformation evolved into real-world harm, and what Trust & Safety teams need to know to stop it.
When War Is a Game: How Video Game Footage Fuels Conflict Misinformation
Video game footage is increasingly misused as fake war content online, shaping conflict narratives and misleading even mainstream outlets. ActiveFence explains how and why, and how trust & safety teams can respond.
Why CISOs Like Me Don’t Sleep in 2025: What You Must Know About Securing GenAI
Discover what really keeps CISOs up at night from our very own Guy Stern, who shares frontline insights into GenAI risk in 2025, exposing hidden vulnerabilities, internal misuse, and how enterprise security must adapt.
Exfiltrating Secrets from LLM Memory: Lessons from the Red Team Trenches
RAG makes AI smarter while also creating new ways for hackers to steal private data. Learn how our Red Team Lab used "memory exfiltration" to trick models into leaking sensitive info through hidden browser requests.
Building Safer AI Agents on Databricks with Alice WonderFence
Discover how Alice and Databricks are partnering to build safer AI agents. Learn how Alice's WonderFence integrates with Databricks’ Mosaic AI Agent Framework to mitigate risks like prompt injection, toxic outputs, and policy violations, ensuring secure, compliant AI deployment at scale
How Scammers Are Abusing GenAI to Impersonate and Manipulate
Scammers are using GenAI to impersonate people, brands, and institutions at scale; fueling fraud, misinformation, and exploitation. Here’s how impersonation abuse works and what AI deployers must do to stay ahead.
How Roleplay and Multi-Turn Prompts Bypass LLM Guardrails
New research shows how roleplaying and multi-turn prompts can bypass LLM moderation and jailbreak protections. This post explains how evasion attacks work, why single-turn filters fail, and how to mitigate real-world risk.
Trusted by security and product teams in the world's most regulated industries
Alice brings years of adversarial intelligence expertise to AI security. We give enterprise teams the coverage that generic guardrails and one-time audits can't match.
Get a Demo