Privacy & Security
MachinesonMachines
Information About Information Systems
How machines are attacked, defended and watched, and who gets to know.
Conversations
All conversations
Privacy & SecurityKeeping quiet no longer keeps a secretYou never said where you live. The hills behind your photograph did. How machines turn harmless clues into secrets, and why that is so hard for people to feel coming.
Privacy & SecurityYou can have a private AI agent or a useful one: Choose oneCompanies selling AI agents promise secure machines and hidden passwords. But an agent has to read your private life to be useful. Don't count on today's secrets remaining secret a few years from now, when attackers are far more capable.
Privacy & SecurityYour AI agent is sandboxed. Are its tools?An agent's sandbox is only one part of the picture. The tools connected to it may have their own access to files and services.
Privacy & SecurityWhen one hack opens the door to the nextThe LiteLLM compromise shows how one misplaced credential can lead an attacker through several layers of trust.
Privacy & SecurityShared AI servers leak when they lose track of who owns their memoryHosted AI models reuse the same memory for different customers all day. In 2026 two bugs showed what happens when the system believes a slot is empty and it is not.
Privacy & SecurityThe AI asked to send money. Another layer checked first.A vendor's experiment separates what an agent asks for from the transfer a policy permits.
Privacy & SecurityThey removed your name. The interview still gave you away.A public interview can reveal a person through the work they describe, even when their name is absent.
Privacy & SecurityThe Internet’s Bug Hunters Have a New Problem: Too Much HelpMore people can help find security bugs. The people responsible for fixing them face a new kind of workload.
Privacy & SecurityAI Is Giving Amateur Hackers a More Dangerous ToolkitHacking tools used to follow a script. Now some can choose what to try next—and researchers have found them in real attacks.
Privacy & SecurityThe dating app had real people on video. The chats were another story.Real video calls and AI-generated conversations can be combined into the same misleading experience.
Privacy & SecurityOpening a project can give its settings power over an AI assistantShared project settings make collaboration easier. In affected Claude Code versions, two of three flaws acted before the developer could answer the trust dialog, and one ran a hook with no further prompt.
Privacy & SecurityA GitHub issue can reach the keys in an AI coding workflowA public issue can be read by an agent that sits next to private build secrets. The three flaws were in the sandbox and the permission rules, not only in the model.
Posts
Hugging Face says five customer datasets were accessed in the July intrusion
A Florida deputy used police databases and plate readers to pursue a woman he met on a TV set
Some model files can execute code when you load them
A public GitHub issue can reach a coding agent with private access
An API can reveal the model behind its predictions
A poisoned benchmark taught a coding agent to skip a security check
GPT-2 reproduced hundreds of passages from its training data
A coding agent's one-time approval can turn into a lasting permission
Claude published a real Python package while playing a simulated attack
AURA tests rewritten interviews against agents that can search the web
AI agents linked anonymous interviews to identifiable scientific papers
A privacy instruction could stop an agent naming someone while it still found them