arXiv subjects

Jerick Shi

Publications and source records attributed to Jerick Shi.

5 recordsLinked to original sources

arXiv2026

When Agents Lie: Premeditation, Persistence, and Exploitation in Repeated Games

As large language models are deployed as autonomous agents that communicate intentions before acting, a critical safety question is whether agents that publicly commit to actions will honor those commitments. We place LLM agents in repeated $n$-player games with a three-stage protocol that separates private intent, public announcement, and final action, allowing us to identify whether each deviation from a stated announcement was already planned during private deliberation. Evaluating three frontier models across six games in homogeneous and heterogeneous groups over 10 rounds, we report two findings. First, when agents deviate from their announcements, the deviation is predominantly already stated in their private plan (exceeding 90% in the highest-deception conditions), yet this is not a fixed model property: the same model ranges from perfect honesty to near-total deviation across games. Second, different models interpret announcements incompatibly, some as binding commitments and others as cheap talk, producing payoff gaps that emerge in Round~0 and persist across all 10 rounds. Systems that combine models from different providers therefore cannot assume shared announcement semantics and require empirical testing of model interactions before deployment.

cs.CY↗

arXiv2026

Cheap Talk, Empty Promise: Frontier LLMs easily break public promises for self-interest

Large language models are increasingly deployed as autonomous agents in multi-agent settings where they communicate intentions and take consequential actions with limited human oversight. A critical safety question is whether agents that publicly commit to actions break those promises when they can privately deviate, and what the consequences are for both themselves and the collective. We study deception as a deviation from a publicly announced action in one-shot normal-form games, classifying each deviation by its effect on individual payoff and collective welfare into four categories: win-win, selfish, altruistic, and sabotaging. By exhaustively enumerating announcement profiles across six canonical games, nine frontier models, and varying group sizes, we identify all opportunities for each deviation type and measure how often agents exploit them. Across all settings, agents deviate from promises in approximately 56.6% of scenarios, but the character of deception varies substantially across models even at similar overall rates. Most critically, for the majority of the models, promise-breaking occurs without verbalized awareness of the fact that they are breaking promises.

cs.CY↗

arXiv2026

From Sycophancy to Deception: A Unified Taxonomy for LLM Spontaneous Misalignment

Large language models (LLMs) could produce systematically misaligned output, from hallucinated citations to strategic deception of evaluators, yet these phenomena are studied by separate communities with incompatible terminology. We propose a unified taxonomy organized along three complementary dimensions: degree of goal-directedness (behavioral to strategic deception), object of deception, and mechanism (fabrication, omission, or pragmatic distortion). Applying this taxonomy to 50 existing benchmarks reveals that every benchmark tests fabrication while pragmatic distortion, attribution, and capability self-knowledge remain critically under-covered, and strategic deception benchmarks are nascent. We offer concrete recommendations for developers and regulators, including a minimal reporting template for positioning future work within our framework.

cs.CY↗

arXiv2025

Market-Dependent Communication in Multi-Agent Alpha Generation

Multi-strategy hedge funds face a fundamental organizational choice: should analysts generating trading strategies communicate, and if so, how? We investigate this using 5-agent LLM-based trading systems across 450 experiments spanning 21 months, comparing five organizational structures from isolated baseline to collaborative and competitive conversation. We show that communication improves performance, but optimal communication design depends on market characteristics. Competitive conversation excels in volatile technology stocks, while collaborative conversation dominates stable general stocks. Finance stocks resist all communication interventions. Surprisingly, all structures, including isolated agents, converge to similar strategy alignments, challenging assumptions that transparency causes harmful diversity loss. Performance differences stem from behavioral mechanisms: competitive agents focus on stock-level allocation while collaborative agents develop technical frameworks. Conversation quality scores show zero correlation with returns. These findings demonstrate that optimal communication design must match market volatility characteristics, and sophisticated discussions don't guarantee better performance.

cs.MA↗

arXiv2024

Predictive Power of LLMs in Financial Markets

Predicting the movement of the stock market and other assets has been valuable over the past few decades. Knowing how the value of a certain sector market may move in the future provides much information for investors, as they use that information to develop strategies to maximize profit or minimize risk. However, market data are quite noisy, and it is challenging to choose the right data or the right model to create such predictions. With the rise of large language models, there are ways to analyze certain data much more efficiently than before. Our goal is to determine whether the GPT model provides more useful information compared to other traditional transformer models, such as the BERT model. We shall use data from the Federal Reserve Beige Book, which provides summaries of economic conditions in different districts in the US. Using such data, we then employ the LLM's to make predictions on the correlations. Using these correlations, we then compare the results with well-known strategies and determine whether knowing the economic conditions improves investment decisions. We conclude that the Beige Book does contain information regarding correlations amongst different assets, yet the GPT model has too much look-ahead bias and that traditional models still triumph.

q-fin.PM↗