Tech executives meet senior US officials for ‘frank discussion’ of AI risks

The executives leading the AI race that has transfixed the tech industry, including the CEOs of OpenAI, Google and Microsoft, met at the White House on Thursday for what a senior US official described as a “frank discussion” about their responsibilities to make sure their systems are safe.

The meeting, which was attended by vice-president Kamala Harris and other senior officials, comes as the administration of Joe Biden scrambles to develop a more concerted response to recent rapid advances in the technology. It also comes days after AI pioneer Geoffrey Hinton sounded one of the most notable warnings about the long-term dangers of developing machines that become more intelligent than humans.

“Many of [the big tech companies] have spoken to their responsibilities, and part of what we want to do is make sure we have a conversation about how they’re going to fulfil those pledges,” the senior administration official said. This person did not outline any further regulations or actions the Biden administration might consider.

Executives invited to the meeting were told it would focus on risks stemming from “current and near-term” development of the technology. But by underlining the “fundamental responsibility to make sure their systems are trustworthy and safe before they’re released or deployed”, it will also help deal with “what’s likely to be much more powerful technologies in the future”, the official said.

After the meeting, Harris said in a statement: “As I shared today with CEOs of companies at the forefront of American AI innovation, the private sector has an ethical, moral, and legal responsibility to ensure the safety and security of their products. And every company must comply with existing laws to protect the American people.”

Biden also dropped by the meeting, according to a report from the event.

Ahead of the meeting, the White House said seven of the largest AI companies had agreed to open their models to a degree of public scrutiny at the annual Def Con hacker convention in August. However, it said the level of openness would be “consistent with responsible disclosure principles”, leaving unclear how much the companies would reveal. OpenAI has refused to release basic technical information about its latest large language model, GPT-4.

The White House also said the Office of Management and Budget would release draft guidelines for public comment this summer governing the federal government’s use of AI.

The tech leaders attending the Thursday meeting were OpenAI chief executive Sam Altman, Microsoft CEO Satya Nadella, and Sundar Pichai, chief executive of Google and Alphabet, along with Dario Amodei, CEO of AI start-up Anthropic. Administration officials at the meeting included Jake Sullivan, the National Security Council director; Lael Brainard, the National Economic Council director; Gina Raimondo, the commerce secretary, and Jeff Zients, the White House chief of staff.

The Biden administration has announced a spate of initiatives around AI in recent months, including releasing a draft AI bill of rights and launching a review in April of what new technical standards are needed to ensure AI systems work as intended, rather than exposing people to unforeseen risks.

Lina Khan, chair of the Federal Trade Commission, sounded one of the administration’s loudest alarms on Thursday. Writing in the New York Times, she said her agency would look at whether existing laws can be used to tackle things such as online scams and privacy violations caused by AI.