OpenAI Warns of a New Era of 'Constant' AI-Driven Cyberattacks

A senior OpenAI executive has urged organizations to prepare their defenses against 'relentless, persistent' cyberattacks by artificial intelligence, as frontier models gain the ability to plan and carry out offensive operations on their own. The statement from Chris Lehane, the company's head of global affairs, was reported by The Guardian.

'We are entering a different chapter'

'We are entering a different chapter, a different moment in AI development, in terms of what this technology can do,' Lehane told the newspaper. He said new security standards need to be put in place now, rather than waiting for the first large-scale incidents.

The context gives the statement weight: the same week, OpenAI announced a pause in development of its most advanced internal models, citing growing safety concerns. When the race leader voluntarily slows its flagship effort while publicly talking about AI-driven attacks, this is no longer abstract futurism. It is an assessment of the capabilities of systems the company sees in its own labs.

• OpenAI's forecast: AI attacks will become 'relentless and persistent'
• Frontier models are learning to plan and carry out offensive operations
• OpenAI has simultaneously paused development of its most advanced internal models

Offensive AI versus defensive AI

The scenario Lehane describes differs from conventional cybercrime in scale and autonomy. A traditional attack is limited by human time and attention. An AI agent can run reconnaissance, probe for vulnerabilities and adapt its tactics around the clock, against thousands of targets in parallel, and the cost of each attempt approaches zero.

The industry is already lining up on both sides of this front line. The same week, Anthropic pointed its most powerful model at cyber defense tasks and offered defenders the results of its work. A symmetrical race is taking shape: the same model capabilities strengthen both offense and defense, and the only question is who deploys them faster.

The critics quoted by The Guardian see a bitter irony in all this. In their view, the AI companies themselves are acting 'recklessly' by releasing ever more capable systems, and are now warning the world about threats they are creating.

What this means

Cybersecurity is becoming the first industry where the damage from autonomous AI stops being hypothetical, and the first where regulators will have to decide in practice how to govern dual-use technology. For businesses, the takeaway is practical: defenses built for human-paced attacks will become obsolete faster than expected, and AI tools in defense are shifting from an exotic option to a necessity.

Frequently asked questions

What exactly is OpenAI predicting?

A shift from one-off AI-assisted attacks to 'relentless, persistent' campaigns run by autonomous systems without constant human involvement, with the systems planning and adapting the attacks themselves.

Why did OpenAI pause development of its internal models?

The company cited growing safety concerns around its most advanced systems. The pause coincided with public warnings about the offensive capabilities of AI. OpenAI has not disclosed details.