Description

A recent investigation by Bitsight TRACE has exposed a stealthy yet highly scalable attack vector leveraging malicious calendar subscriptions on iOS and macOS devices. Over 4 million devices are unknowingly syncing daily with 390 sinkholed domains previously used to distribute spam, phishing links, and potentially malicious payloads. Attackers exploit deceptive pop-ups or fake CAPTCHA prompts to trick users into subscribing to calendars. Once subscribed, the device continues polling the domain indefinitely. If the domain later expires and is re-registered by threat actors, they gain instant, persistent access to push harmful events to millions of devices without requiring further user interaction. This threat is amplified by an inherent trust users place in system-generated notifications, making calendar-based attacks harder to detect than traditional phishing attempts. Bitsight highlighted a significant escalation: the emergence of AI-driven “Promptware.” As AI assistants like Google Gemini integrate deeply with personal data, malicious actors can embed hidden prompts or jailbreak commands within calendar event descriptions. When a user asks their assistant to summarize their schedule, the AI may inadvertently execute the embedded malicious instructions, potentially leading to unauthorized actions such as sending emails, exfiltrating data, or deleting content—without the user ever realizing the source. To mitigate these risks, users and organizations should routinely audit device calendar subscriptions and remove any unrecognized or unnecessary entries. Implement domain-monitoring practices to detect expired calendar hosts that could be re-registered by attackers. Mobile device management (MDM) solutions should enforce restrictions on unauthorized calendar subscriptions. Finally, AI assistant vendors must strengthen safeguards against indirect prompt injection to ensure that personal data sources cannot silently trigger harmful commands.