… For example, PromptArmor demonstrated that the link preview feature in messaging apps like Telegram and Discord can be turned into a data exfiltration pathway through indirect prompt injection. …
… This was a gold mine for cyber threat actors leveraging prompt injection techniques , as they could quietly exfiltrate session tokens to remote servers. …
… Prompt injection isn't solved, and a well-engineered project is not the same as a well-configured deployment. …
… They leak in CI logs, or they get scraped in supply-chain attacks, or, sometimes, they're just sitting in plain sight. …
… Finally, the model also shows significant improvement in agentic safety, meaning it's a lot better at recognizing and refusing prompt injection attacks when you're using it as an agent. …
… Prompt injection is a real and documented attack vector - malicious instructions hidden inside a webpage or an email can manipulate what an agent does on your behalf without you realising it. …
… LLMs are vulnerable to prompt injection, where you can hijack the chain to put your own instructions in. …
… Sending it all back to me, unprompted, is a spectacular way to break that trust. …