Live Markets-Can AI be kept safe without handing the keys to big tech?

A key flashpoint in the debate is self-learning AI. Supporters of tighter controls warn that advanced systems could eventually improve themselves, evade safeguards or act in ways that humans do not fully understand. As a result, they argue for stricter oversight of powerful AI models and the computing infrastructure behind them.

Critics counter that concentrating AI development in the hands of a few technology giants creates its own risks. They believe open-source development and broad community scrutiny can help identify vulnerabilities more quickly and create a more resilient defense against misuse.

Palumbo's conclusion is that AI regulation should be proportional to the risks involved. Lower-risk applications should face lighter requirements, while high-stakes uses such as healthcare, finance and law enforcement should be subject to more rigorous oversight. The most advanced AI systems, meanwhile, warrant the strongest safeguards. The goal, he says, is to protect the public without creating regulatory barriers that cement the dominance of today's largest technology companies.

Philip Palumbo, founder, CEO and chief investment officer of Palumbo Wealth Management, argues in a note out late Friday that the debate over AI guardrails is less about choosing between safety and innovation and more about finding the right balance between the two.

Guardrails such as safety testing, privacy protections and human oversight are important because the risks associated with advanced AI are real. However, Palumbo notes that the biggest advocates for stricter regulation are often the largest technology companies, which are best equipped to handle the costs of compliance. That raises concerns that regulations could unintentionally make it harder for startups, university researchers and open-source developers to compete.