Customer agent bots with inadequate AI pose a growing risk to the brand Clio

Last year, a grieving traveler asked Air Canada’s chatbot about bereavement fees. The bot invented a refund policy that didn’t exist. The customer acted accordingly, the airline ended up in court and the story it went viral. The court rejected Air Canada’s argument that its chatbot was a “separate legal entity” responsible for its actions and ordered the airline to pay damages.

That incident is now a cautionary tale for every brand scaling AI into customer communications. And new research from customer communications platform Sinch suggests this is not an isolated case.

According to one study, around 74% of companies have already been forced to roll back a distributed AI agent due to governance failures Sinreport “The paradox of AI production” by ch. Here’s the twist: Companies with the most mature guardrails—those that invested most in compliance, security protocols, and oversight—backtracked at an even higher rate of 81%. Teams that do more to prevent failure fail more often, not less.

“If governance were the solution, more mature teams would go back less, not more,” said Daniel Morris, chief product officer at Sinch. “Engineering teams spend most of their time building and maintaining security systems instead of focusing on improving the customer experience. This is the guardrail fee that slows organizations down.”

The impact of the guardrail tax

For marketing teams, the guardrail fee has a direct cost. Every hour spent by technicians rebuilding security infrastructure is an hour not spent improving customer experience and generating revenue.

Air Canada is not alone. A car dealership chatbot agreed to sell a Chevy Tahoe for $1 after a prank. A AI support bot at coding startup Cursor invented a non-existent access policy, triggering a wave of customer cancellations. A the delivery company’s bot he swore at a client and wrote a poem disparaging his employer. Each incident went viral. Each one damaged a brand. And each helps explain Sinch’s finding that three out of four companies have already reinstated a distributed AI agent.

Sinch surveyed 2,527 business decision makers across 10 countries and six industries. The results that matter most to marketers:

• 62% of companies already have AI communication agents in production, and 88% plan to deploy one within 12 months. The pressure to take sides is intense.
• 74% were forced to roll back a deployed agent due to governance failures. Three out of four marketing organizations have already felt the pain of an AI implementation that had to be cancelled.
• 84% of teams spend at least half of their design time rebuilding their security infrastructure from scratch. This is engineering capability that could go towards personalization, channel expansion, and campaign optimization.
• When an AI agent fails, 35% of the impact impacts the support queue. Almost the same percentage, 34%, depends on brand perception, and this aspect is more difficult to correct.

The study found that infrastructure quality was the single strongest predictor of implementation success, surpassing model choice, team size, and budget. However, most organizations say their current vendor falls short in at least one critical area.

AI-powered customer communication agents manage customer conversations at scale: chatbots on websites, voice agents in contact centers, SMS and email auto-responders, and omnichannel platforms that route and respond across channels. They range from simple FAQ bots to sophisticated agents that authenticate users, process transactions, and personalize responses based on customer history.

Sinch’s research tracks agents already in production, not pilots or internal experiments. These are systems that marketers rely on every day, where failure means frustrated customers, longer wait times, and brand damage that spreads within minutes.

Choosing the wrong base is the real risk

Jayashree Iyangar, global head of CX data and artificial intelligence at HGS, a digital experience company, said the results match what she sees on the ground. Marketers are past the pilot phase, he noted, and the real challenge lies in operations.

“The key question is how AI can be seamlessly orchestrated across multiple channels, not whether it can be deployed in just one,” Iyangar said.

He pointed out that the risk profile varies significantly depending on the use case. A marketing chatbot fumbling with a promotional offer carries less weight than a service agent mishandling a sensitive billing issue. “Human oversight remains central in service environments where the risk of negative impact on customers is greatest,” he said. “This is also where we see more cases of AI rollback.”

His take on infrastructure echoes Sinch’s seminal finding. “A significant portion of effort is spent on building and maintaining security systems rather than improving the customer experience,” he said. It is seeing organizations consolidate around centralized AI governance teams that manage trust, compliance and security separately from the AI ​​use cases themselves.

Three moves marketers can make now

For marketing teams, the study points to three practical moves.

1. Let infrastructure drive your vendor decision. Infrastructure quality predicts implementation success more than any other variable in the Sinch data. When evaluating vendors, ask about guardrail engineering, multi-channel orchestration, and the extent to which your team will absorb the security burden. The right platform handles most of the security work, so your team can focus on the customer experience.
2. Plan the guardrail tax into your schedule. Security systems are not a one-time installation cost. They consume ongoing engineering resources that would otherwise be dedicated to CX improvements. Budget for that reality from the start rather than watching the timeline slip as rollbacks occur.
3. Push for a separate governance function. Iyangar’s observation about centralized AI governance teams aligns directly with the data. Keeping AI use cases and governance engineering separate reduces overall costs. Marketing should have no security infrastructure. It should partner with a dedicated governance function that manages trust, compliance and security, allowing marketing to focus on work that directly touches customers.