STORY · FORSKNING_
The most dangerous AI hacking techniques still require humans in the loop
Security researcher James Kettle presented findings at the Black Hat conference about agentic AI's ability to develop new hacking techniques. He concluded that AI alone has limited ability to autonomously develop entirely new attack methods, but becomes extremely powerful when combined with human guidance. Kettle uncovered a new vulnerability type called Shared-Parser Confusion through collaboration between humans and AI models.
WHY IT MATTERS
The finding demonstrates both the potential and limitations of AI in cybersecurity, and illustrates that the most effective attacks will likely require human-AI collaboration rather than purely autonomous AI systems. This has implications for both defensive and offensive security going forward.
SOURCES
MACHINE-GENERATED SUMMARY This summary is written by machine from the sources below. We sort and explain — but we are a way into the field, not the final word. Check the source when something matters to you.