JPMorgan Chase CEO Jamie Dimon has identified cyber risk as one of the bank’s largest threats, warning that artificial intelligence is creating vulnerabilities that attackers may exploit at greater speed and scale. In JPMorgan’s April 2026 shareholder letter, Dimon said AI would worsen cyber risk but described the dangers as manageable if companies, regulators and governments prepare. He also said JPMorgan was investing heavily in protection and remaining vigilant.
Anthropic’s own testing found that Mythos Preview could identify complex vulnerabilities, turn them into exploit primitives and combine those primitives into complete attack chains.
That threat is expanding alongside corporate use of chatbots, AI agents and other applications that can access internal systems and sensitive data. SentinelOne’s Singularity platform is designed to protect cloud environments, identities, endpoints and other parts of an enterprise network, while also providing tools for investigation and recovery after an attack. Its Purple AI assistant can investigate threats and take defensive action, and its Prompt Security product is aimed at prompt-injection attacks that attempt to trick AI applications into disclosing protected information.
SentinelOne’s latest reported quarter showed growing demand for those capabilities. More than half of total recurring revenue came from non-endpoint products, including data, cloud and AI security.
The company remained unprofitable under generally accepted accounting principles, although its second-quarter non-GAAP operating margin improved to 10% from 2% a year earlier.
The investment case therefore depends on whether rapid adoption of AI-security products can continue translating into recurring revenue and eventually durable profits. Dimon’s warning does not establish that SentinelOne will benefit, but it highlights the security problem that the company is positioning its products to address. Anthropic’s testing confirms that advanced models can perform increasingly capable vulnerability research, while SentinelOne’s results show that customers are already buying tools built for security in an AI-enabled environment.