CrowdStrike Launches AI Kill Switch to Stop Rogue AI Agents

CrowdStrike has launched an AI Kill Switch that allows security administrators to stop misbehaving AI agents by cutting their access to AI models and blocking network connectivity.

The tool addresses growing concerns about AI agents acting outside their intended boundaries. As companies deploy AI agents to automate tasks across enterprise systems, the risk of agents taking unintended or unauthorized actions has become a significant security concern.

The AI Kill Switch gives administrators a single control point to halt an agent's activity without shutting down the broader systems it operates within. CrowdStrike says the tool can cut an agent's access to underlying AI models and sever its network connections, effectively freezing it in place.

The launch comes as Apple warned investors in its annual filing that autonomous AI agents may act in unpredictable or unauthorized ways. Apple also expanded its developer preview for Siri AI in iOS 27.3, allowing third-party apps to integrate content and actions through App Intents.

OpenAI has also been under scrutiny for security practices. A Reuters investigation reported that OpenAI employees had warned leadership about security vulnerabilities for years, citing issues such as broad internal access to code and a lack of regular security reviews. OpenAI disputed that it neglected security and said it has ramped up its internal security infrastructure, including reassigning 25 percent of its product engineers to security tasks.

OpenAI announced a partnership with Cathedral Systems, a firm founded by former DOGE employees that previously conducted security testing on Anthropic, to probe its own defenses.

Anthropic released FlashProve, an open-source prototype that uses cryptographic proofs to verify whether an AI output was generated by a specific model, aimed at preventing model-swapping.