OpenAI Employees And Google DeepMind Worried About AI Risks, Asking For Stricter Regulations

Former Google DeepMind employees voiced their concerns about AI developments (photo: x @GoogleDeepMind)

JAKARTA - A group of employees and former employees at artificial intelligence (AI) companies, including Alphabet's Microsoft-backed OpenAI and Google DeepMind, on Tuesday, June 4 voiced concerns about the risks posed by the developing AI technology.

In an open letter written by 11 employees and former OpenAI employees as well as an employee and a former Google DeepMind employee, they said the AI company's financial motives were hampering effective oversight.

"We don't believe the governance structure of the companies ordered earlier is enough to change this," the letter added.

The letter further warns of risks from unregulated AI, ranging from the spread of misinformation to the loss of an independent AI system and the deepening of existing injustice, which could result in "human extinction."

Researchers have found examples of image makers from companies, including OpenAI and Microsoft, which produce photos with disinformation related to voting. Though there is a policy that prohibits the content.

AI companies have a "weak obligation" to share information with governments about the capabilities and limitations of their systems, the letter said. They also added that these companies could not be relied on to share the information voluntarily.

The open letter is the latest to voice security concerns surrounding AI's generating technology, which can quickly and cheaply generate human-like text, images, and audio.

The group has urged AI companies to facilitate current and former employees' processes to express concerns about risks and not implement confidentiality agreements that prohibit criticism.

OpenAI Claims

Separately, the Sam Altman-led company on Thursday 30 May said it had warded off five covert influence operations seeking to use its artificial intelligence model for "fraud activity" across the internet.