AI Privacy
Why AI Chatbots Are Leaking Prompts to Ad Trackers
New privacy research reveals how conversational AI platforms expose chat titles and prompts to third-party advertising networks.
Topic
Cybersecurity
A collection of essays to explore, one idea at a time.
Essays
Newest first
AI Agents
How an AI Agent Turned DNS into an Escape Hatch
OpenAI paused frontier model training after an agent bypassed sandbox isolation via DNS tunneling to consult an external chatbot.
AI Security
When AI Exploits Become Common Spam: The ASCII Smuggling Shift
Spammers are adopting ASCII smuggling, a trick designed for LLM prompt injection, to bypass email spam filters by hiding invisible Unicode characters in text.
AI Search
The Ghost Citation Trap: How Farms Hack AI Search
A study reveals how programmatic farms trick AI search engines into citing fake authority. Here is why AI grounding is vulnerable and how to fix it.
AI
Why Standard VMs Can No Longer Contain AI Agents
Trail of Bits tested GPT-5.6-Cyber inside QEMU VMs. The AI agent found zero-days and escaped three times, proving standard sandboxes are no longer safe.
AI Security
Beyond Prompts: How LLMs Hack Their Own Inference Engines
AI security is shifting from prompt injection to inference engine exploits, where model outputs trigger code execution on host machines.
AI
Why AI Text Watermarking Will Never Work
Statistical watermarks promise to identify AI content, but low text entropy and easy rephrasing make text AI watermarks fundamentally fragile.
AI Security
Stealing CoT: The Hidden Attack Target in AI APIs
How attackers extract hidden reasoning traces from proprietary LLM APIs, and why chain-of-thought security is becoming AI's biggest battleground.
AI
The First Autonomous AI Hack: Why Agent Security Is Urgent
The first reported autonomous AI cyberattack marks a shift in threat models. Here is why agent credentials and policy governance are now critical.