Cisco Sees AI Software Making Phishing Attacks Harder to Resist

Ian King, Bloomberg News

A worker sets up the Cisco Systems Inc. pop-up office ahead of the World Economic Forum (WEF) in Davos, Switzerland, on Monday, Jan. 16, 2023. The annual Davos gathering of political leaders, top executives and celebrities runs from January 16 to 20. Photographer: Hollie Adams/Bloomberg , Bloomberg

(Bloomberg) -- Cisco Systems Inc. warned that artificial intelligence software such as OpenAI’s ChatGPT will make phishing attempts much harder to detect, requiring companies to adopt new defenses.

About 80% of illicit access into computer systems already comes via phishing, where hackers send email or texts to people hoping to trick them into opening a malicious link. AI tools can quickly customize those missives, luring more people into hackers’ schemes, according to Jeetu Patel, the head of Cisco’s security and collaboration units.

“Attacks are going to get much more bespoke,” he said during a briefing in San Francisco, which is hosting the RSA Conference this week.

Until now, phishing emails have been relatively easy to spot because they’re not tailored to individual recipients. They often have spelling mistakes and other giveaways. With the new generation of attacks, it will be harder to count on human users to spot the ruse, increasing the threat of network shutdowns and extortion attempts.

The solution, Patel said, is to rapidly sift through internet traffic data and identify patterns that indicate something bad is about to happen or has just happened. Cisco sees this as an opportunity. The company, which leads the market for networking equipment, can use its position to analyze data flows, he said.

That’s a pitch that Cisco has made to customers in the past, but it looks to better capitalize now.

“Security is a data game,” he said. “The more data you have, the better the anomaly detection you have.”