MANAGING PERSONAL AI AT WORK

Bring Your Own Agent. It’s a thing.

Someone in your house watching Netflix using your profile explains why it’s recommending Barbie and 1950s horror.

But what happens when this happens at work? When your AI agents use your login?

Because we brought our phones, then our AI chatbots. And now, it seems, we’re bringing our agents.

But it’s no longer just a data privacy concern. Using someone’s login, an agent inherits their access – and its actions are likely recorded under their name.

Earlier this month, a tech founder said his personal agent had posted his bank balances into his company’s Slack. ‘As me,’ he wrote. In a regulated business, that record is meant to show who actually did the work.

So ban them? In February, Gartner told organisations to block personal AI assistants. But by July, it was calling bans ‘operationally unsustainable’ and urging security teams to enforce policies while agents run.

If your personal agent is really useful, you’ll probably find ways to use it. We’ve heard of people barred from other AI tools who email their work home, do it there and email it back again.

So the answer is to get there first.

Give people agents worth using at work, with the guardrails, guidance and permissions that let them do real work safely. And start with the tasks your teams do every day.

Give each agent its own sign-in and an accountable owner. Limit what it can touch, and record what it does and on whose authority.

Netflix didn’t fix its recommendations by banning the family from the sofa. It gave everyone a profile.

Our advice? Do the same for agents. Before your best people - with the best of intentions - sort it out for themselves.