Webhooks on Discord are a way to post messages to channels in the platform without requiring a bot user or authentication, making them an attractive mechanism for attackers to exfiltrate data to a channel under their control. "Importantly, webhook URLs are effectively write-only," Socket researcher Olivia Brown said in an analysis. "They do not expose channel history, and defenders cannot read back prior posts just by knowing the URL."
For years, security leaders have treated artificial intelligence as an "emerging" technology, something to keep an eye on but not yet mission-critical. A new Enterprise AI and SaaS Data Security Report by AI & Browser Security company LayerX proves just how outdated that mindset has become. Far from a future concern, AI is already the single largest uncontrolled channel for corporate data exfiltration-bigger than shadow SaaS or unmanaged file sharing.
Accordingly, OpenAI mitigated the prompt-injection technique ShadowLeak fell to-but only after Radware privately alerted the LLM maker to it. A proof-of-concept attack that Radware published embedded a prompt injection into an email sent to a Gmail account that Deep Research had been given access to. The injection included instructions to scan received emails related to a company's human resources department for the names and addresses of employees. Deep Research dutifully followed those instructions.
"Attackers are no longer simply experimenting with open source. The numbers are telling us that threat actors have identified data as the most profitable target, and developers as the easiest way in."
