SAN FRANCISCO: Alphabet Inc (GOOGL.O) is cautioning staff about how they use chatbots, including its own Bard, while also marketing the program globally, according to four people familiar with the situation.
According to the people, and confirmed by the company, the Google parent has cautioned staff not to enter private materials into AI chatbots, citing long-standing policy on information security. Chatbots, such as Bard and ChatGPT, are human-sounding programs that use generative artificial intelligence to communicate with users and respond to a variety of inquiries. Human reviewers may read the conversations, and researchers discovered that similar AI might duplicate the material it received during training, posing a risk of data leakage.
According to some of the sources, Alphabet also warned its engineers to avoid using straight computer code generated by chatbots. When asked for comment, the company stated that while Bard can provide undesirable code changes, it nevertheless helps engineers. Google has stated that it intends to be open about the limitations of its technology.
The concerns demonstrate Google’s desire to avert financial damage from the software it released in competition with ChatGPT. Billions of dollars in investment and untold advertising and cloud revenue from new AI programs are at play in Google’s race against ChatGPT’s backers, OpenAI and Microsoft Corp (MSFT.O).
According to Reuters, a rising number of businesses around the world have put guardrails in place for AI chatbots, including Samsung (005930. KS), Amazon.com (AMZN.O), and Deutsche Bank (DBKGn.DE). Apple (AAPL.O), which did not respond for comment, is also said to have done so. According to a survey of roughly 12,000 respondents, including from prominent U.S.-based organisations, 43% of professionals were utilising ChatGPT or other AI technologies as of January, often without informing their supervisors.
According to Insider, Google advised staff testing Bard before to its introduction in February not to provide it with internal information. Google is now bringing Bard to more than 180 countries and 40 languages as a creative tool, and its cautions extend to its code suggestions. Google told Reuters that it has had lengthy discussions with Ireland’s Data Protection Commission and is responding to regulators’ questions, following a Politico report Tuesday that the company was postponing Bard’s EU launch this week pending further information about the chatbot’s impact on privacy.
