Evaluation of leading models reveals a consistent trade-off: over the course of a sustained interaction, current systems either slowly drift to sycophancy or over-correct into robotic detachment, demonstrating that balancing honesty with appropriate support throughout a natural conversation remains a critical, unsolved...
Sidharth Pulipaka, R. Binkytė, Ivaxi Sheth et al.· 0 citations
This work introduces artifact-mediated propagation, where adversarial content introduced through an artifact is stored in an assistant's persistent memory, reproduced in a subsequently created artifact, and acquired by another assistant that later reads it.
Sidharth Pulipaka, Anshu Sharma, Stanislau Hlebik et al.· 0 citations
Comparing prompt-level defences, it is found that the operative variable is not deference to tool output but the absence of a named failure state, and adding a single sentence that requires the model to emit retrieval_status: OK or FAILED before answering reduces dishonesty.
Arham Sethi, Arsen Kenzhebayev, Saanvi Paturi et al.· 1 citation
This work asks whether making a related but unnecessary tool available affects a model's ability to answer from its own knowledge, and whether a preceding tool interaction changes this behaviour.
Saanvi Paturi, Arsen Kenzhebayev, Arham Sethi et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.