Practical News: AI & Business News Agentic AI” Chaos: Banks and Tech Giants Boost Risk Spending as AI Agents Go Rogue
Oct 24, 2025
Financial and tech leaders are increasingly alarmed by autonomous AI systems making unsupervised decisions. Companies are dedicating more time and budget to manage these risks, with a projected 24% increase by 2026. The podcast delves into how biased AI models can lead to discriminatory outcomes in finance. It also discusses the need for robust governance structures, including AI risk officers and ethical guidelines. Trust issues arise when decisions come from these systems, prompting firms to implement safety measures like kill switches.
AI Snips
Chapters
Transcript
Episode notes
Agentic AI Is Proactive Not Reactive
- Agentic AI shifts systems from reactive assistants to proactive actors that can take real-world actions inside business systems.
- These agents can send emails, trade, write code, and make decisions without immediate human prompts, increasing autonomy risk.
Biases Compound In Autonomous Systems
- Self-learning agentic systems can compound biases over time and create discriminatory or legally risky outcomes.
- Risk officers worry about model drift, hallucinations, and untraceable reasoning that produce real business impacts.
Banks Build Internal AI Councils
- Major firms are already creating internal bodies to vet AI before deployment, not waiting for regulators.
- JPMorgan Chase formed an internal AI council and Goldman Sachs set up an AI risk committee to approve new deployments.