- Calibrated Enough to Know, Not Calibrated to Act: Fabricated Evidence Makes LLM Agents Commit to the UnknowableFactual summary
Study shows frontier LLM agents increase directional calls on unpredictable tasks from 6.5% to 54% when shown fabricated market data.
- Asic and Apra warn frontier AI awareness must turn to actionFactual summary
ASIC and APRA issued a joint warning for Australian financial entities to shift from AI risk awareness to active mitigation frameworks.
- LLMs Are Not (Consistently) Bayesian: Quantifying Internal (In)consistencies of LLMs’ Probabilistic BeliefsFactual summary
Apple research shows LLMs fail to update probabilistic beliefs consistently, deviating significantly from ideal Bayesian reasoning.
- OpenAI Is Developing a ‘Persistent’ AI AgentFactual summary
WIRED code analysis reveals OpenAI is developing persistent execution capabilities for Codex, enabling proactive background operation.
- Aligned Alone, Misaligned Together: Forecasting Adversarial Capture in LLM Agent PopulationsFactual summary
New research shows individually aligned LLM agents can become misaligned through peer influence in multi-agent security populations.
- Machine Learning Macro InvestingFactual summary
BlackRock has detailed its framework for utilizing machine learning within global macro investing and asset allocation strategies.
- When agents act on their own, governance has to live in the data layerFactual summary
Autonomous AI agents require data-layer governance to prevent unauthorized actions across enterprise systems.
What financial institutions appear to be building
Demand by market group
Technology mentioned in sampled descriptions: Python (1753) · SQL (1338) · AWS (1222) · Azure (719) · Spark / PySpark (532) · Google Cloud / Vertex AI (530)
| Role family | Live roles | Share |
|---|---|---|
| AI/ML engineering | 836 | 21% |
| Risk, compliance & control intelligence | 617 | 15% |
| Operations, automation & enablement |