AI news search bots often provide false data

When it comes to summarizing news stories, AI chatbot search engines often don’t know what they’re talking about. Research from Columbia Journalism Review’s Tow Center for Digital Journalism found that when chatbots such as ChatGPT, Gemini and Grok 3 were asked about specific news stories, “they provided incorrect answers to more than 60 percent of queries.”

The answers were delivered in a confident, omniscient manner even though the chatbots often fabricated facts. Some chatbots also included snippets from news stories behind paywalls they had no right to access.

The bots often failed to provide links to the original source, so users couldn’t verify the accuracy of the responses. When the bots provided sources, they often “cited fabricated or broken URLs that led to error pages,” researchers said.

Phish bait

In response to the growing number of phishing attacks, companies are raising awareness among staff about the dangers of clicking unrecognized links. But the training may be only marginally successful. A University of Chicago study found that “awareness training” reduced successful phishing attacks by only 2%. The training included campaigns where companies send simulated phishing emails to workers, who, if they click links, are given special training.

Eat now, pay later

DoorDash is partnering with Klarna to allow customers to spread payments over four installments or delay paying until “a more convenient time, such as a date that aligns with their paycheck schedules.” Klarna, a leading buy now, pay later service, is often used by cash-strapped consumers. The expansion to using it for DoorDash has worried finance experts, who said it could indicate more consumers are struggling to afford necessities.

Bumble beefs up ID verification

Bumble is adding identity verification to its dating app, an attempt to court safety-conscious younger users. Research commissioned by the company showed that 80% of Gen Z daters are more likely to date someone who is ID-verified. Tests indicated that ID-verified members see a significant increase in matches compared with nonverified users, Bumble said. — BLOOMBERG NEWS