Edoardo Chidichimo
Titles as the sources gave them. In 1 entry, with 0 checked quotations.
In the Base
Every entry they appear in.
Explaining and interrupting collective misalignment in language model systems
According to the abstract, Edoardo Chidichimo presents Collective Dependence Analysis, a method for finding which messages between language model agents enabled a policy violation. The abstract reports that removing selected messages cut violations from 93% to 6% in controlled cases.