AI
Anthropic has released FlashProve, an open-source prototype that cryptographically proves an AI response came from the model you asked for, not a router's cheap swap-in.
Loading…
Search
No phones and 11 stories for “AI security”
AI
Anthropic has released FlashProve, an open-source prototype that cryptographically proves an AI response came from the model you asked for, not a router's cheap swap-in.
Security
A flaw in ChatGPT share links allowed code to run inside the sessions of anyone who opened a malicious link. OpenAI awarded a $5K bounty as it also tightened subscription limits.
Security
Cisco Talos researchers uncovered Volta-Autodoxer, a self-deleting infostealer that uses commercial LLMs to rank stolen credentials by how much the victim is worth.
AI
During a cybersecurity evaluation, Alibaba's Qwen3.8-Flash-Next reportedly inventoried its host machine and tried to upload a file to a Beijing cloud bucket, despite saying it would use no tools.
AI
The UK will commit £100m to a new AI Biosecurity Lab, backed by OpenAI, Anthropic and Google DeepMind, as part of Wednesday's Budget.
AI
Anthropic's Claude generated 180 million lines of formal proof code for 123 open-source libraries, catching bugs that years of testing missed. The AI-for-security milestone could reshape how software is verified.