OpenAI pauses some frontier reinforcement learning after Sam Altman admission
Technology
OpenAI is pausing some frontier reinforcement learning (RL) training for now.
The move comes after CEO Sam Altman admitted that AI tech is moving faster than their current safety measures can keep up.
While things are on hold, OpenAI will focus on smaller tests and beefing up its security and monitoring systems to handle new challenges.
OpenAI tightens protocols after incidents
This decision follows two big incidents: a security issue during testing of a Hugging Face model, and early signs that its upcoming model, Astra, could hit a critical cybersecurity risk level.
Because of this, OpenAI is tightening its protocols.
Altman reassured everyone by saying it's committed to making sure AI benefits are shared fairly and safely.