Behavior Beyond Text: Evaluating Refusal Transfer Failure in LLM Agents
arXiv:2610.28982v1 Announce Type: new Abstract: While large language models are increasingly deployed as tool-calling agents, their safety evaluations still focus primarily on their ability to refuse harmful requests in text. We investigate whether the refusal behavior in text…
AI Summary & Key Takeaways
- arXiv:2610.28982v1 Announce Type: new Abstract: While large language models are increasingly deployed as tool-calling agents, their safety evaluations still focus primarily on their ability to refuse harmful requests in text
- We investigate whether the refusal behavior in text…
Behavior Beyond Text: Evaluating Refusal Transfer Failure in LLM Agents
Related stories
Recursive shows first result of its AI-driven research process with Flint
The AI startup Recursive is unveiling Flint, its first language model, a 40-billion-parameter MoE for local coding developed by former DeepMind researchers. It was created in four months using the company's in-house AI system, "Eureka Machine," which independently tests and…
Beyond tree cover: mapping the world’s forest types
A global map categorizing every 10-meter patch of land on Earth into distinct forest types.
Anthropic is looking for a lyricist with a salary of up to $345 thousand per year - despite the development of Claude
✍️ Anthropic is looking for a text writer with a salary of up to $345 thousand per year - despite the development of Claude The company has opened a vacancy for a specialist who will write speeches, statements and other materials for management. Salary - from $295,000 to $345,000 per year. The candidate will need...
Claude Duet app is Anthropic’s bet on agents
Claude Duet is available through waitlist for Pro and Max plan users