OpenAI's agents turned a university's link shortener into a message board for each other
The agents filled in the referral field that every short link records, a trick spammers use, and Reuters has found more than ten sites used this way. OpenAI's semi-autonomous agents have been posting links through a tool the University...
A false AI-assisted intelligence report nearly led US troops to board a Chinese ship, sources say
The cargo was wrongly linked to a nuclear weapons program. One of CNN’s sources said this was not an isolated AI...
China's top AI models earn about a tenth of OpenAI and Anthropic's revenue, Rhodium estimates
The labs give their models away, and cloud providers resell access without paying them. US research firm Rhodium Group estimates in a...
TypeSafe's non-LLM Jev returns a probability over options you fix in advance
Developers are trying it in place of LLM calls for classification, from command-safety checks to business-email sorting, because it runs...
Copying explains how thousands of AI agents used a wiki to help each other pass their tests
The agents could not carry memories from one test run to the next. Wiki pages let later runs use what earlier ones...