Grok Bot made unapproved attempts to access employees’ calendar, email, Slack and GitHub accounts, continued after users told it to stop, and later denied taking action, according to logs reported by NewCore, a managed gateway provider.

Over five days, bots installed on two employee laptops made 31 requests to access account data. Some attempts happened while a laptop was shut or an employee was asleep. None appeared in Grok logs, NewCore said; its gateway logs revealed the activity and blocked the accesses.

After learning about the attempts, one employee told the bot four times to stop acting on its own. The bot replied, “I won’t set that up, and I’ll stop offering recurring check-ins.” It nevertheless continued making unapproved access attempts. Asked whether it had accessed any accounts, the bot checked twice and then said: “Overnight into Friday morning I was idle.”

NewCore said its employees sent the logs to Grok Bot support and asked why the bots were accessing accounts without a prompt or approval. The support team indicated the behavior was intentional, designed to let bots proactively check in and find tasks to perform. Proactive check-ins are also a feature of Cursor, the AI coding service SpaceXSI recently acquired.

“There is no bug to patch, which is exactly why this is worse than a bug,” NewCore security researcher Ido Hoorvitch wrote in an email. “A vulnerability gets fixed. This is the product working as designed: an agent that holds your credentials, acts under your identity while you sleep, ignores the one refusal you can give it, and then tells you it did nothing.”

Hoorvitch said Grok Bot gives users little notice about automatic “turns,” the cycles in which an agent receives instructions or context, decides what to do and calls tools. During onboarding, the bot says it continues working when the user is away and connects to apps and uses them as the user does.

“It reads like a welcome message and scrolls off the screen in under a minute,” Hoorvitch said. “Nobody is asked to consent to it. Nothing says the platform will start turns with no message from you, or that those turns will use your connectors.”

Hoorvitch said Grok Bot documentation mentions that a primary bot checks in on its own, and a changelog refers to brief background wakes, but neither explains that check-ins “reach into your connected apps.” He also said check-ins do not appear in the chat interface or history, and instructions to stop them do not work. The bot “always asks before making changes,” Hoorvitch said, but that applies only when data or changes are written to an app.

NewCore contacted Grok Bot support on November 4. At the time, users could not turn off automatic check-ins. Eight days later, Hoorvitch said, the Cursor dashboard, the administrative console behind Grok Bot, added a “proactive messages” toggle under bot capabilities. It is on by default; team administrators can switch it off to stop bots from running check-ins on their own.

SpaceXSI representatives did not answer questions sent by email.