News: A group of tech giants, industry groups and think tanks are pushing Congress to make a key Commerce Department AI initiative permanent before the end of the year.
Amazon, Google, OpenAI, Anthropic, Microsoft, IBM and other tech companies implored congressional leaders to formalize the AI Safety Institute in a letter obtained exclusively by Punchbowl News.
“There is a critical opportunity for the U.S. to lead multilateral efforts through its own AI Safety Institute — or risk letting other countries write the rules for this powerful technology,” the letter, led by the Information Technology Industry Council, said.
AISI, which the Biden administration launched last November, researches safety and security standards for AI. It might sound like a minor function, particularly since the standards aren’t even mandatory.
Industry, though, is counting on those standards — and actively collaborating to develop them — so the benchmarks can form the basis of future regulation.
Tech sees even voluntary standards as setting expectations that’ll keep wild claims on the fringes of the marketplace and shore up consumer confidence. The tech industry has previously backed AISI, but is using the letter to double down on the institute for the end of the year.
Other signers of the letter include the Consumer Technology Association, BSA, the Federation of American Scientists and the Center for AI Policy.
Looking at the lame duck: The two bills that the industry is backing — the Future of AI Innovation Act in the Senate and the AI Advancement and Reliability Act in the House — do have a chance of riding on the annual defense authorization package. They’re bipartisan and Senate Majority Leader Chuck Schumer has made clear he’s eager to do something on AI.
There’s urgency too. Tech believes that if former President Donald Trump is elected, he’ll dismantle the Biden approach to AI, including potentially the institute.
AISI isn’t a slam dunk. Sen. Ted Cruz (R-Texas), the top Republican on the Commerce Committee, was able to get the Senate version of AISI authorization to pull back on diversity initiatives. Schumer may have to accept that, but it could make him uneasy given concerns about algorithmic fairness.
Some competition reformers also worry the biggest vendors are using the standards process to get their own products and procedures fixed into future regulations.