The plan of Amodei from Anthropic was simple: Don’t allow anybody to use as open source their AI models, since they might remove piece of trash, called Constitution, demand data and money to do the development of the models only in Anthropic, declare AI so dangerous, that American government should block fast development of any smaller competitors, and you have your monopoly at the end. So far this doesn’t work:
"WASHINGTON -- President Trump postponed the signing of an executive order that would have given the government more oversight over the artificial-intelligence industry, saying he didn't want to take any action that would slow down the U.S. in the AI race.
Hours before a scheduled signing event Thursday that was set to include industry executives, Trump said in the Oval Office that the order would have asked AI companies to preview models with the federal government. Such a move would set back the U.S. in its competition with China, which AI analysts say the U.S. is winning, he said.
"I really thought that could have been a blocker, and I want to make sure it's not," he said, citing the number of jobs the technology is creating. "I don't want to do anything that will get in the way of that lead."
The move is the latest example of the White House being divided on AI policy. Administration officials have been trying to bolster cybersecurity following the release of powerful models including Anthropic's Mythos, but some AI advisers fear the steps could slow down the development of new AI tools, creating two camps within the administration that are quarreling over the tech agenda.
Trump's move came hours after California Gov. Gavin Newsom, a Democrat, issued an executive order mandating that the state prepare policies to address the job losses and economic disruptions caused by AI, which are contributing to a growing backlash against the technology.
Trump's decision is a win for business-focused advisers, including venture capitalist David Sacks, who has pushed for an industry-friendly approach to AI. Sacks and others have argued that the economic benefits of AI models outweigh the risks.
A senior administration official recently suggested that the White House was considering potential AI rules comparable to the system used by the Food and Drug Administration to approve medicines.
Sacks said recently on the "All-In" podcast he co-hosts that Trump and senior officials didn't want an FDA-like process for AI and that Silicon Valley was alarmed by the concept because it could slow down model development. "There's a very visceral negative reaction here because we know how damaging that would be to innovation," he said. Sacks co-chairs the President's Council of Advisors on Science and Technology and was previously AI and crypto czar.
The administration is considering more oversight after Amodei noise with Mythos worried top White House officials and business leaders who fear the models are so good at identifying software vulnerabilities that they are capable of cyberattacks. OpenAI is previewing a similar model to customers and, like Anthropic, is restricting access with the government's input.
The postponed executive order would have asked companies to voluntarily preview advanced models with the government. It would have asked national-security and cyber officials to work with agency heads and top tech companies to address software vulnerabilities identified by models including Mythos.
The goal was to protect entities such as small-town hospitals and banks that don't have full access to the best technology for sophisticated cyber defenses. Their protection is a priority for administration officials, including Vice President JD Vance and Treasury Secretary Scott Bessent.
"We're trying to balance that safety against innovation," Vance said at a press briefing Tuesday. "We think that we've got the right balance here in the Trump administration, but it's something we're going to have to keep on working on."
A recent survey from the conservative-leaning Institute for Family Studies, a think tank that supports more AI guardrails, showed that more than 80% of U.S. registered voters support AI testing before models are released.
The executive order would have been reminiscent of actions taken during the Biden administration, which tried to compel top AI companies to notify the government when developing powerful models. Trump, a Republican, unwound many of the prior administration's actions on AI.
Trump has proposed a framework that the administration has said includes adequate oversight of companies and protection for consumers, though many Democrats and Republicans said the proposal isn't robust enough to become law.” [1]
1. President Delays Order on AI, Fearing U.S. Would Fall Behind. Ramkumar, Amrith. Wall Street Journal, Eastern edition; New York, N.Y.. 22 May 2026: A1.
Komentarų nėra:
Rašyti komentarą