English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Recent advances in Language Model (LM) agents and tool use, exemplified byapplications like ChatGPT Plugins, enable a rich set of capabilities but alsoamplify potential risks - such as leaking private data or causing financiallosses. Identifying these risks is labor-intensive, necessitating implementingthe tools, manually setting up the environment for each test scenario, andfinding risky cases. As tools and agents become more complex, the high cost oftesting these agents will make it increasingly difficult to find high-stakes,long-tailed risks. To address these challenges, we introduce ToolEmu: aframework that uses an LM to emulate tool execution and enables the testing ofLM agents against a diverse range of tools and scenarios, without manualinstantiation. Alongside the emulator, we develop an LM-based automatic safetyevaluator that examines agent failures and quantifies associated risks. We testboth the tool emulator and evaluator through human evaluation and find that68.8% of failures identified with ToolEmu would be valid real-world agentfailures. Using our curated initial benchmark consisting of 36 high-stakestools and 144 test cases, we provide a quantitative risk analysis of current LMagents and identify numerous failures with potentially severe outcomes.Notably, even the safest LM agent exhibits such failures 23.9% of the timeaccording to our evaluator, underscoring the need to develop safer LM agentsfor real-world deployment.

Identifying the Risks of LM Agents with an LM-Emulated Sandbox