Google is privately warning its staff not to reveal sensitive information to chatbots, while publicly touting the future of its own AI bot.

According to Reuters, who cite “four people familiar with the matter”, Google has advised its employees not to enter any confidential information into AI chatbots, including to its own Bard bot, and to avoid direct use of any computer code that chatbots spit out.

Google confirmed that Bard can “make undesired code suggestions”, which is concerning confirming a recent GitHub survey of 500 US-based enterprise developers found 92 per cent of them are already using AI coding tools both in and outside of work.

Google also told staffers not to enter any sensitive company information into Bard, despite launching Bard in more than 180 countries and in 40 languages.

The only nod to these potential dangers came when Google updated its privacy notice on June 1 to include the phrase: “Don’t include confidential or sensitive information in your Bard conversations.”