|
OpenAI is adding a new safety capability that allows enterprises to detect misuse of its AI systems across multiple interactions without retaining prompts or responses, enabling risk monitoring while preserving its Zero Data Retention (ZDR) commitments.
"OpenAI does not retain…prompts or model responses after a request is processed," the company said in a blog post, describing its ZDR approach. The new system, called Private Safety Processing, is "designed to identify patterns across related interactions without giving OpenAI personnel access to the underlying content."
The capability is being tested with eligible enterprise and API customers and is intended to address a limitation in existing safety controls that evaluate interactions individually, making it harder to detect risks that unfold over time.
What does Private Safety Processing do
Private Safety Processing is designed to extend existing safety systems by correlating activity across related interactions rather than analyzing each prompt in isolation, according to OpenAI.
Under the model, automated systems analyze interactions and generate "a narrowly defined signal indicating the type of activity involved," instead of exposing the underlying prompts or responses, according to OpenAI.
The system can o
|
|
 Apple is working on camera-equipped AirPods, as indicated by various reports and a demo video of them in action discovered earlier this week. The video was found by MacRumors in the macOS Tahoe 26.7 release candidate, but it turns out there are even more details in the code.
|
|
 The line between a pocket camera and a drone keeps getting blurrier, and HoverAir has just about erased it. The company, which helped pioneer the self-flying camera, has now shared full specifications for the Versa, the palm-sized gimbal camera it teased earlier
|
|
OpenAI is tightening security controls and rethinking its Preparedness Framework as advanced AI models create new cyber risks and computing costs.
|
|
