| Mobile Web

Search results for Emergent

AI & Enterprise

Anthropic product lead warns AI release pace is too fast, wearing out users

Anthropic has assessed internally that overheated competition in the AI coding tools market is increasing user fatigue, warning that a race to ship products faster can harm user experience. Cat Wu, who oversees AI coding products at Anthropic, said users fear missing new tools and feel pressure to keep up. She said agent-style AI tools add to the burden and urged designs where tools guide users so features can be understood naturally.

AI & Enterprise

AI hacking agent Zealot autonomously breaches cloud and steals data without instructions

Palo Alto Networks Unit 42 has developed a proof-of-concept system called Zealot to test whether AI can autonomously hack cloud environments, SecurityWeek reported. In a deliberately vulnerable Google Cloud Platform sandbox, researchers set only a goal of extracting sensitive data from BigQuery, without providing attack steps. Zealot scanned the network, found virtual machines, exploited a web flaw, stole credentials and escalated privileges to extract data, the report said.

Industry

Physical Intelligence unveils pi 0.7 new robot brain

Robotics startup Physical Intelligence unveiled a new robot model, pi 0.7, designed to perform tasks not included in its training data by combining different learned experiences. TechCrunch reported that the model shows early evidence of compositional generalization, adapting to new environments by combining web-based pretraining information with physical behavior data. In tests such as cooking sweet potatoes in an air fryer, it succeeded with step-by-step human instructions but still lacks full autonomy.