U.S. workers are turning to ChatGPT to help with basic tasks despite fears that have led employers such as Apple, Microsoft, Google, Samsung, Verizon and many others to ban its use on work computers and devices, a Reuters/Ipsos poll found.

Some 28% of 2,625 adult respondents to the online poll on artificial intelligence (AI) conducted between July 11 and July 17 said they regularly use ChatGPT at work, while only 22% said their employers allowed such external tools.

About 10% of those polled said their bosses have banned external AI tools, while 25% did not know whether their company permitted the use of the technology.

Data from Similarweb interprets the data a bit differently than the numbers from Reuters and Iposos. Web traffic to some of the AI chatbots and search engines has dropped in use during the summer months.

One newcomer, Perplexity.AI, has increased in popularity in July, while other AI-based chatbots and search engines such as OpenAI’s ChatGPT saw traffic decline, according to Similarweb data released this week.

Internet traffic to Perplexity.AI — a search engine and chatbot also powered by the same technology as ChatGPT — rose 9.6% in June and 11.2% in July. Bard, Google’s chatbot, stagnated in June but rebounded in July, rising 34.5%. 

Similarweb reported that Perplexity.AI has mitigated some of the problems with generative AI — for example, by providing footnotes that link back to verifiable sources of the data it is presenting and a system for filtering out “hallucination” effects where AI engines have been known to “bluff” their way through answers with contrived facts. 

Another newcomer, Character.ai, saw traffic drop 32.4% in June, but rebounded in July with an 8.3% increase.

Workers are turning to ChatGPT to help with basic tasks despite fears that have led firms including Apple, Microsoft, Google, Samsung and Verizon to ban its use on work computers and devices, a Reuters/Ipsos poll found.