AI in contact centers: The hidden threat to customer service and security

Share on LinkedIn Share on LinkedIn

Artificial intelligence (AI) is reshaping the contact center industry, promising faster service, improved customer insights, and enhanced automation. Yet, beneath the surface, serious challenges persist. Despite AI-driven analytics, many businesses still struggle to understand why customers are calling. At the same time, AI-powered chatbots and virtual assistants introduce security vulnerabilities, leaving companies exposed to risks like prompt hacking. Is AI in contact centers a game-changer—or a growing problem?

The Data Paradox: more information, less understanding

Why are customers calling? AI Agents still don’t have the answer

Contact centers process millions of interactions daily, capturing data on call times, wait durations, sentiment analysis, customer history, and issue categories. Yet, even with these insights, many businesses fail to uncover the true intent behind customer calls.Consider these common scenarios:

  • If customers frequently call about refunds, is the issue with the product itself, the return process, or unclear policies?
  • If customers endure long wait times, is it due to staffing shortages, inefficient routing, or high call complexity?

Raw data alone doesn’t tell the full story. Without deeper insights, businesses risk misdiagnosing problems and implementing ineffective solutions.

Why AI Agents struggle to deliver actionable insights 

AI Agents generate reports, not understanding

AI-powered analytics tools can generate reports instantly, but they lack contextual intelligence. A report showing refund-related calls with the lowest Net Promoter Score (NPS) doesn’t clarify:

  • Whether refund dissatisfaction is expected or fixable
  • If a poor NPS score correlates with other service issues
  • Whether customer frustration stems from policy, process, or overall experience 

Fragmented legacy systems hinder AI Agent effectiveness

Many enterprises still operate with outdated, fragmented technologies. When AI Agents pull data from incomplete or disconnected sources, their conclusions can be misleading, leading to:

  • Partial customer insights that don’t reflect the full journey
  • Wasted resources fixing the wrong problems
  • Decision-making based on misleading trends

For AI Agents to deliver meaningful insights, they need unified, high-quality data—something many contact centers still lack.

AI security risks: The threat of prompt hacking

While AI Agents struggle to understand customer intent, they face an even bigger problem: security vulnerabilities. What is prompt hacking?

When AI chatbots are powered by large language models (LLMs), they become vulnerable to prompt hacking—where bad actors manipulate chatbot responses to extract sensitive information or bypass restrictions.

Hackers can:

  • Trick AI Agents into revealing confidential data
  • Manipulate bots into executing unauthorized transactions
  • Exploit AI loopholes to spread misinformation

Why Voice AI is more secure

Unlike text-based AI Agents, voice AI is harder to manipulate because:

  • It is difficult to make many voice calls in a short period of time trying different prompts
  • It requires more context to process queries
  • It can incorporate voice authentication for security

 While text-based AI Agents improve efficiency, businesses cannot afford to ignore the security risks that come with them.

Scaling AI: Are contact centers ready?

Companies like Genesys, Teneo and AWS demonstrate how AI can scale customer service. However, scaling AI Agents without proper data intelligence and security can lead to:

  • Unreliable customer insights
  • AI-driven miscommunication
  • Increased vulnerability to cyber threats

The challenge isn’t just deploying AI Agents—it’s ensuring they enhance customer experience rather than undermining it.

The human element: Why AI alone isn’t enough

AI Agents can process vast amounts of data, but some lack emotional intelligence and human judgment (In the world of AI this is translated to low accuracy). Customers still value personal interactions, particularly when dealing with complex or sensitive issues. AI can assist, but it shouldn’t replace human representatives entirely. Instead, businesses should focus on integrating AI Agents with human support to strike the right balance between efficiency and empathy.

Customer service leaders should:

  • Use LLM´s to build but deterministic AI-models to run in order avoid unnecessary security risks
  • Implement escalation paths for the 20-30% of calls where AI can’t support the customer and seamlessly transfers cases to human agents when necessary.
  • Continuously refine AI models based on real customer feedback and frontline employee insights.

Final thoughts: AI Agents in customer service—risk or reward?

AI Agents have the potential to revolutionize contact centers, but without real customer insight and strong security measures, they can become liabilities rather than assets. To succeed, businesses must:

  • Prioritize actionable insights over raw data
  • Unify fragmented channels and systems to improve AI Agent effectiveness
  • Invest in AI security to prevent prompt hacking risks
  • Balance AI-driven automation with human support for a seamless customer experience

In the race to automate, companies must not lose sight of what truly matters—understanding their customers and safeguarding their trust. Only then can AI truly become a game-changer for contact centers.

Share on LinkedIn Share on LinkedIn

Marie Angselius
Marie Angselius-Schönbeck is Chief Impact Officer and Chief Marketing Officer at Teneo.ai, a company in voice first Agentic AI. In 2019, she founded Women in AI by Amelia, a global initiative to help close the gender gap in STEM. She has worked in th Conversational AI-industry for 7 years.

ADD YOUR COMMENT

Please use comments to add value to the discussion. Maximum one link to an educational blog post or article. We will NOT PUBLISH brief comments like "good post," comments that mainly promote links, or comments with links to companies, products, or services.

Please enter your comment!
Please enter your name here