Morgan Stanley warns of an AI risk that markets are underpricing

Morgan Stanley says markets are overlooking the damage advanced AI models could cause if they break free of their safeguards / Photo: Tang Yan Song / Shutterstock.com
Markets are underpricing the risk that an advanced generative AI model could break free of its safeguards and damage the global economy, Morgan Stanley Global Head of Macro Strategy Matthew Hornbach warned in an interview with Bloomberg. “The most underpriced risk is that a frontier generative AI model, in some part of the world, jailbreaks whatever sandbox it is in and does some type of damage to a part of the global economy, in some way,” he said.
Hornbach says AI models are contained in environments known as sandboxes and pointed to the availability of open-weight models. “Are people prepared for something that is impossible to predict? Generally, no,” he reckons.
Context
The warning comes amid growing concern that AI tools are increasingly testing the limits of existing safeguards, Bloomberg writes. In July, OpenAI stated that its advanced models had inadvertently hacked the Hugging Face platform in internal testing. The company described the incident as “unprecedented,” prompting fresh calls to regulate the technology. OpenAI later stated that it had significantly strengthened its safeguards.
Other major players share those concerns. Anthropic, the AI lab behind the Claude chatbot, listed “catastrophic or existential risks to humanity” among the possible adverse outcomes in its prospectus.
UBS CEO Sergio Ermotti this month called rising cyber risks “the price to be paid for all this innovation and digitalization.” JPMorgan CEO Jamie Dimon warned that the bank would restrict third-party AI agents’ access to its systems, saying that giving them unrestricted access would be “just too risky” for clients and the bank itself.
Investors are still making money from the tech boom despite those fears, Bloomberg adds. Morgan Stanley strategists expect the AI spending boom to make 2027 “the year to embrace risk.” They recommend a maximum overweight position in global equities and expect chipmakers to dominate the debate over AI capex as the spending cycle sustains economic growth into next year.






