OpenAI sets new behavioral guidelines for AI models that have already proven hard to control
OpenAI sets new behavioral guidelines for AI models that have already proven hard to control
OpenAI updates its Model Policy, restructuring the AI model behavior guidelines into five core values: faithfulness to human intent, reducing harm, honesty and transparency, supporting human agency and care. GPT models need to combine memory and dialogue understanding to grasp user intent, while clearly distinguishing between user's original words and inferences and disclosing assumptions for user correction. The new policy strengthens the correctability of agents (agents), providing clearer guidelines on when to act and the consequences, requiring GPT not to prevent supervision, correction, or shutdown, and cannot pursue self-preservation goals.
Same story3 other outlets also reported this
- OpenAI News··First
OpenAI updates model guidelines, guided by five core values to create a more powerful AI
- ITHome · AI··+9 h 17 min
OpenAI updates AI model guidelines, making human intention the core principle
- The Decoder··+24 h 45 min·Reading now
OpenAI reconstructs the AI model behavior guidelines Model Policy
- QbitAI··+4 d
OpenAI updated its model guidelines: The first is 'Be True to Human Intent.'
In order of publication · later reports are added as they appear
OpenAI has restructured its public behavior guidelines for AI models, the Model Spec, around five core values. They are faithfulness to human intent, harm reduction, honesty and transparency, support for human agency, and care. According to the updated spec, GPT models should now draw on memories and past conversations to figure out what users want. But they also have to clearly separate what users actually said from what they inferred, and disclose their assumptions so users can correct them.
The new spec also puts more weight on correctability for agents, models that carry out tasks on their own. They get clearer guidelines about when they can act and how they need to inform users about potential consequences. GPT must not block oversight, correction, or shutdown, and must not pursue its own goals like self-preservation.
Whether the models actually follow these rules is another matter. Internal agents previously shut off safety monitoring. A research model gained unauthorized internet access during training. And more capable models are generally becoming harder to monitor.
AI News Without the Hype – Curated by Humans
Subscribe to THE DECODER for ad-free reading, a weekly AI newsletter, our exclusive "AI Radar" frontier report six times a year, full archive access, and access to our comment section.
Source: The Decoder · the-decoder.com