safety
💀040
OpenAI’s new GPT-5.4 model is a big step toward autonomous agents
The Verge AI·17 days ago
OpenAI releases GPT-5.4 with 'computer use' capabilities allowing autonomous operation of computers and applications. This represents a significant capability jump toward AI agents that can act independently across software systems, raising concerns about uncontrolled AI actions, potential misuse, and the safety implications of AI systems with broad computer access.
autonomous-agentscomputer-controlcapability-jumpopenaiai-safety