With AI, hackers are like tigers with wings

On July 18, the American cybersecurity company Proofpoint (headquartered in Sunnyvale, California) released a statistical report in May showing that more than 80% of email scams with identifiable senders were sent to Japanese-speaking addresses.

This shows that advances in artificial intelligence (AI) have allowed for more natural language to be used.

Yukimi Sota, an expert at Proofpoint's Japan branch, said that of the 770 million phishing emails sent globally in May, 240 million contained sender data and 81.4% of them targeted Japanese speakers.

Phishing emails were previously easy to spot due to their unnatural wording, but advances in AI have helped generate natural sentences, allowing hackers to overcome language barriers, Sota said.

Proofpoint's report shows that the volume of phishing emails began to increase sharply from February 2022. Before 2025, there were 100-200 million such emails sent per month, but that number skyrocketed to more than 500 million emails/month in 2025.

Many phishing emails are sent from addresses that impersonate brokerage firms. They direct recipients to fake websites that are used to steal personal information such as email addresses and passwords, allowing hackers to take over accounts.

If a company's emails and security information are stolen, attackers can gain unauthorized access to internal communication systems, from which they can send additional phishing emails.

Faced with this worrying reality, Ms. Sota called on Japanese companies to strengthen cybersecurity measures such as adopting multi-factor authentication.