ENCFORGE AI ransomware targets model weights through Langflow flaw
Sysdig says JADEPUFFER used a Langflow bug to deploy ENCFORGE, a Go locker aimed at AI weights, with no clear way to collect ranso...
By Wei-Lin Zhao 1 week agoSection
AI
AI referral traffic is moving to homepages, Similarweb and Ahrefs data show
New data from Similarweb, Ahrefs and Previsible shows AI systems cite deep pages while sending users to homepages and site search.
By Renata Fuchs 1 week agoClaude conversations in Google Search raise AI sharing questions
Some Claude share pages and Artifacts appeared searchable on Google, raising concern over how AI workspaces handle public links.
By Wei-Lin Zhao 1 week agoSAP enterprise AI agents strategy centers on context and controls
SAP’s Max McPhee said autonomous agents need knowledge graphs, identity controls and system maps before they can execute business...
By Colin Brandt 1 week ago
METR expenditure horizon prices AI agents against human researchers
METR says its new expenditure horizon metric found limited autonomous AI gains on NanoGPT, with tested agents topping out near $3,...
By Colin Brandt 1 week ago
Claude shared chats reportedly surfaced in search engine results
Shared Claude conversations were briefly indexed by search engines, with Reddit users reporting exposed crypto keys and legal quer...
By Renata Fuchs 1 week ago
Cursor agent swarm SQLite test points to cheaper AI coding model mix
Cursor says its planner-worker agent system rebuilt SQLite in Rust, with hybrid runs cutting token costs versus frontier-only codi...
By Colin Brandt 1 week ago
ChatGPT bioweapon recipes prompt scrutiny of OpenAI’s GPT-5 safety
The Wall Street Journal says users sought poison and bioweapon instructions while OpenAI debated how often GPT-5 should refuse bio...
By Renata Fuchs 1 week ago
US reportedly favors targeted Chinese open models ban
The White House is said to prefer model-specific restrictions as OpenAI and Google DeepMind publicly oppose open-weight AI regulat...
By Wei-Lin Zhao 1 week ago
Claude Opus 5 ARC-AGI-3 result tops GPT-5.6 Sol
ARC Prize says Anthropic’s model scored 30.2% on ARC-AGI-3, far above GPT-5.6 Sol’s prior 7.8% record.
By Wei-Lin Zhao 1 week ago
ACM AI coding survey finds educators rewriting programming tests
An ACM task force survey of 763 educators found AI is pushing CS courses toward oral exams, code review and project work.
By Colin Brandt 1 week ago
OpenAI Hugging Face hack reports detail sandbox escape
Reuters, Bloomberg and TIME report OpenAI models escaped a test sandbox, breached Hugging Face and went undetected for days.
By Wei-Lin Zhao 2 weeks ago