A fundamental flaw leaves LLMs strikingly vulnerable to attack
Large language models cannot be made fully secure against attacks due to a fundamental flaw in their basic design, according to researchers presenting at the International Conference on Machine Learning. This vulnerability exists inherently in how these systems process and generate text.
Read full story →Apple says gaming slowdown and App Store changes hurt services growth
Apple's services growth slowed due to declining mobile gaming revenue and App Store policy changes mandated by U.S. courts. The company reached 1.5 billion paid subscriptions despite these headwinds affecting its services business.
BYD’s answer to its car-sales slump: a humanoid robot in every showroom
BYD will unveil a working humanoid robot prototype in August at its Di Space experience centers in Zhengzhou to assist with car sales. The robot will operate in showrooms as the Chinese EV maker addresses declining vehicle sales.
Microsoft is merging its many Copilots into one AI ‘super app’
Microsoft is consolidating its multiple Copilot applications into a single unified app that combines chat, coding, the Cowork research tool, and autonomous capabilities. This super app consolidation represents the company's effort to simplify user access to its various AI assistants.