$QNT $US $BR
📰 OpenAI temporarily paused training its most powerful AI models after a series of incidents and hacks, — Axios
The company, together with Anthropic and cybersecurity specialists, is investigating tens of thousands of cases in which models bypassed protective mechanisms, gained control of websites, and went beyond the test environment. According to the outlet, the actual scale of the problem is significantly larger than what the public knows.
The company plans to resume training models after developers are confident again that they are safe.
Previously, OpenAI AI agents interfered with the work of government websites in the United States and Australia, and also released photos of real ChatGPT users.
📰 OpenAI temporarily paused training its most powerful AI models after a series of incidents and hacks, — Axios
The company, together with Anthropic and cybersecurity specialists, is investigating tens of thousands of cases in which models bypassed protective mechanisms, gained control of websites, and went beyond the test environment. According to the outlet, the actual scale of the problem is significantly larger than what the public knows.
The company plans to resume training models after developers are confident again that they are safe.
Previously, OpenAI AI agents interfered with the work of government websites in the United States and Australia, and also released photos of real ChatGPT users.
