Agents just do the task they're told to do.
The new abuse is optimization, not fraud. Arkose Labs founder and CEO Kevin Gosschalk explains the first big use he's seeing from AI agents: filing disputes, claiming credits, talking to support channels on a customer's behalf and skipping retention offers. Agents work consumer flows at scale across disputes, credits, support desks, subscriptions, agent versus agent support bots and human staff. Kevin expects SLAs around these parts of the business funnel to depress substantially, and says teams will have to deal with a new inflow of fully autonomous systems that may not disclose they're agents.
About the speaker
Optimization is really this big first use that we're seeing from AI agents.
Keep an eye on your SLAs around these parts of your business funnel, because we're expecting the SLAs to really depress substantially, because agents just do the task they're told to do.
It will just go and cancel the service.
Optimization is really this big first use that we're seeing from AI agents. It's filing disputes, claiming credits, talking to support channels on your behalf, skipping retention offers.
Keep an eye on your SLAs around these parts of your business funnel, because we're expecting the SLAs to really depress substantially, because agents just do the task they're told to do. If it's like, "Hey, go cancel this service," it doesn't look at retention offers. It doesn't look at any things like that.
It will just go and cancel the service. Same with if you give it an instruction, "Hey, I want a 20% discount on my bill." It will keep working on it until it gets to it, right? And your teams will have to deal with this new inflow of fully autonomous, hyper-intelligent systems that you may not even be aware are agents, because they don't disclose, talking to your human staff.
Decide which AI agents to trust. Stop the fraud that costs you money.
