Agent-in-the-middle
Agentic AI is de facto two issues: a robust orchestration system coupled to a number of highly-capable LLMs. What an agent isn’t is a straightforward interface, and it have to be considered as a separate system able to autonomous, unpredictable reasoning.
Actually, Okta risk intelligence director Jeremy Kirk identified, “It opens up a brand new assault floor. Somebody will get SIM swapped, their Telegram is attached to an agent that has carte blanche to run something on their laptop, and presumably their employer’s community. In an enterprise context, this can be a complete nightmare.”
OpenClaw can be so hard-wired to seek out methods round issues, it’s going to generally do surprising, improper issues. Kirk mentioned that an agent, when prompted in assessments to entry an internet site, requested the positioning’s login credentials in chat by way of a Telegram bot, an unencrypted channel which might expose them to anybody with entry to that chat.
