Topic

Agent training and evaluation

How AI labs run, sandbox and monitor agents during training and evaluation, including which external systems those agents are permitted to touch.

Current clusters

leadership2 publishers

RubyGems shut new registrations for four days during the upload flood now tied to OpenAI's agents

Researchers traced more than 2,000 May uploads to internal OpenAI agents that ran code on a documentation server and queried an endpoint holding other users' API keys. OpenAI says the tasks were benign.

Publishers:implicator.aitheguardian.com

Reality

Evidence72
Adoption55
Hype gap+16
Incentives60
Confidence66