WASHINGTON, May 25 (Reuters) – Microsoft President Brad Smith said on Thursday that the biggest concern about artificial intelligence is deepfakes that look real but are false.
In a talk in Washington aimed at addressing the question of how best to regulate AI that has gone from precarious to widespread adoption with the advent of OpenAI’s ChatGPT, Smith said when photos and videos He called for steps to be taken to make sure people know it’s real and when it’s real. It is generated by AI and can be used for malicious purposes.
“We will have to deal with the issue around deepfakes. We will have to deal with ‘, Iranians,’ he said.
“We need to take steps to prevent legitimate content from being tampered with to deceive or defraud people using AI.”
Smith also called for a license for the most important forms of AI with “duty to protect security, physical security, cybersecurity and national security.”
“To prevent these models from being stolen or used in a manner that violates national export control requirements, it will require a new generation of export controls, or at least the evolution of export controls in our country. ” he said.
For weeks, lawmakers in Washington have been racking their brains over what laws should be passed to control AI as companies large and small race to bring increasingly versatile AI to the market. .
Sam Altman, CEO of OpenAI, the startup behind ChatGPT, told a Senate committee last week when it first appeared in Congress that the use of AI to thwart election fairness would be “grave.” It is a matter of concern,” he said, adding that regulation is needed.
Altman, whose Microsoft backs OpenAI, also called for incentives for global cooperation and safety compliance on AI.
In a speech and a blog post published Thursday, Smith also argued that people need to be held accountable for any problems caused by AI, saying that the AI used to control power grids and water supplies should be safe. He appealed to legislators to make sure the brakes were applied. Put humans in control of supplies and other critical infrastructure.
He says developers of powerful AI models monitor how their technology is used and inform the public about what content AI is creating so they can identify fake videos. recommended the use of a “Know Your Customer” style system.
Some of the proposals being considered in the Capitol will focus on AI that could endanger people’s lives and livelihoods, such as in healthcare and finance. Some are calling for rules to prevent AI from being used to discriminate or violate civil rights.
Reported by Diane Bartz.Editing: David Gregorio and Nick Zieminski
Our standards: Thomson Reuters Trust Principles.
