An endless blitz of automated deepfakes, disinformation and cyberattacks from America’s adversaries: this is what keeps the CIA’s director for artificial intelligence up at night.

“What AI sometimes enables in these spaces is the ability to do it at a scale and speed that hasn’t been possible,” Lakshmi Raman said at the POLITICO AI & Tech Summit on Wednesday. “It’s much more available, and much easier for people.”

The CIA is especially keeping an eye on China’s AI program, seen as a key threat to national security, Raman said. She noted one of the biggest concerns with China is the country’s undemocratic approach to leveraging the powerful emerging tech.

“They are growing every which way,” Raman said of China.

But even as AI gives adversaries more tools against the U.S., it also provides the CIA with critical new capabilities, Raman said. She said AI helps CIA agents pore through reams of data and identify patterns and trends that would be difficult or impossible to spot otherwise. By peeling away low-level tasks, the technology gives intelligence officers more time to focus on strategic work, she said.

Raman commented on how the CIA is also developing an internal AI chatbot meant to help with research and writing. But she said to keep innovating, the agency will need to keep recruiting. “We need the people who can do this kind of work,” Raman said. “And the truth is the demand far exceeds the supply.”