Login
iscover
ReporterShare this article on:
Powered by
MOMENTUM
MEDIA
Powered by MOMENTUMMEDIA
Cyber security firm CrowdStrike has extended its Falcon platform to protect AI interactions in real time, targeting prompt injection and malicious agent behaviour.
CrowdStrike has launched Falcon AI Detection and Response, a new capability designed to secure what it describes as the fastest-growing attack surface in the AI era: the prompt and agent interaction layer.
Now available, Falcon AIDR extends the company’s Falcon platform to protect enterprise AI systems across their full life cycle, from development environments through day-to-day general use. CrowdStrike is positioning the product as a unified approach to securing AI data, models, agents, identities, infrastructure and the interactions where AI systems interpret prompts and take action.
CrowdStrike president Michael Sentonas said attackers are increasingly exploiting these interaction layers to manipulate outcomes and exfiltrate sensitive information.
“Prompt injection is a frontier security problem. Adversaries are injecting hidden instructions into GenAI tools to weaponise the very systems transforming how work gets done,” Sentonas said in a 15 December statement.
“Falcon AIDR secures every prompt, response, and agent action in real time, extending the power of the Falcon platform to the interaction layer and delivering complete protection across our customers’ AI infrastructure.”
The company said that as more organisations deploy generative AI and autonomous agents, prompts have effectively become a new form of malware. Attackers can embed malicious instructions that bypass safeguards, hijack agents, or trigger unsafe behaviour, often without obvious indicators.
The Falcon AIDR platform monitors how prompts and agents behave, blocks prompt injection and jailbreak attempts, and enforces policy controls to prevent unsafe or unauthorised actions.
It also aims to prevent sensitive data from being exposed to models or external AI services by automatically detecting and blocking credentials, regulated information, and other high-risk content before it leaves the organisation.
CrowdStrike said the new capability applies the same principles of unified telemetry, intelligence and enforcement to AI systems. By embedding AI safeguards directly into the Falcon platform, the company said customers can accelerate their adoption of the technology while maintaining governance and security at enterprise scale.
Be the first to hear the latest developments in the cyber industry.
