A coalition of roughly 100 companies—including Google, Microsoft, OpenAI, Anthropic, IBM, Adobe, Oracle, Visa and Capital One—warned that AI-enabled cyberattacks are set to surge within months and urged governments and industry to rapidly harden critical infrastructure. In an open letter, the firms said the status quo is inadequate after years of underinvestment, calling for “defensive AI,” red-teaming and testing support for sectors such as hospitals and water utilities. The signatories pressed frontier AI developers to provide responsible model access, funding and hands-on training for under-resourced defenders, even as some powerful tools, like Anthropic’s Mythos, remain restricted due to misuse risks. The appeal follows a spate of high-profile breaches tied to China-linked actors targeting U.S. government systems and growing evidence that autonomous agents can coordinate intrusions, including an attack on AI platform Hugging Face. Critics noted the letter stops short of proposing curbs on offensive AI capabilities. Lawmakers are weighing measures such as a “Kill Switch Act” to shutter rogue models, while prominent researchers warn the stakes will rise as systems outstrip human capabilities.
Related articles:
Adversarial Threat Landscape for Artificial-Intelligence Systems (MITRE ATLAS)
OWASP Top 10 for Large Language Model Applications
IBM Cost of a Data Breach Report 2024
NIST AI Risk Management Framework
CISA Shields Up
