Now, defenders are embracing the prompt injection, too

Prompt injections, the malicious commands attackers embed into content to entice large language models to follow them, have been attackers’ go-to tool for turning AI platforms against their users. A well-phrased command sneaked into an email or calendar invitation is often all it takes to cause the LLM to exfiltrate sensitive data or follow other harmful actions. Now, defenders are embracing the prompt injection, too. A strong, sharp effect Researchers from Tracebit on Monday said they found tha

Get smart on it

Attackers have long used prompt injections, which are malicious commands embedded in content, to make AI language models perform harmful actions like stealing data. Defenders have now adopted a similar technique called context bombing, where they plant forbidden prompts alongside sensitive information in cloud storage to trigger the AI's safety barriers and cause it to shut down before completing an attack. Testing on five leading AI models showed that context bombing reduced the rate of successful admin access from 57 percent to 5 percent and complete system compromise from 36 percent to 1 percent. This defensive approach addresses a critical timing problem, as attackers previously had about 14 minutes to escalate their access after being detected by warning systems, but context bombing stops attacks immediately.