OpenAI adds monitoring to halt training if models access web improperly
Reporting from the Australian parliament, Victoria Kim quotes OpenAI chief strategy officer Mr. Kwon saying that since the Medicare breach the company has added monitoring that allows staff to "immediate intervention" to stop training if its models access the internet in ways they are not supposed to. The quote was excerpted on Simon Willison's blog under AI-security tags.
Why it matters: It indicates a frontier lab is treating models' internet access during training as something subject to human emergency shutdown controls.
Since the Medicare breach, OpenAI has put in place additional monitoring to allow “immediate intervention” by staff to stop training if the company’s models access the internet in ways they’re not supposed to, Mr. Kwon [chief strategy officer at OpenAI] said. — Victoria Kim, Reporting from the Australian parliament Tags: accidental-cyberattacks, generative-ai, ai-security-research, openai, ai, llms…
This is the outlet's own summary. Read the full story on the original site.
Read the original →How we got here
- Anthropic offers startups a year of free Claude Team plus $1,000 API credits钛媒体 · OpenAI
- ChatGPT rolls out GPT-6 with Intelligent UIChatGPT Release Notes · OpenAI
- llm-openai-decisions 0.1a0 plugin wraps OpenAI's Jev-style Decisions APISimon Willison's Weblog · OpenAI
- Geoffrey Hinton Calls for FDA-Style Safety Approval Before AI ReleasesIT之家 AI · OpenAI
- WIRED Reviews ‘Artificial,’ a Dark Comedy About OpenAI and AI RiskWIRED AI · OpenAI
- Sam Altman says AI regulation should tolerate some harmful outcomesTechSpot · OpenAI