特朗普拒绝 Anthropic、OpenAI 与 xAI CEO 放缓 AI 的呼吁

US President Donald Trump attends the Irish Open at the Trump International Golf Links & Hotel in Doonbeg, Co Clare, during his trip to the Republic of Ireland. Picture date: Sunday September 13, 2026. (Photo by Liam McBurney/PA Images via Getty Images) (Liam McBurney - PA Images via Getty Images) President Trump dismissed a rare bipartisan call from the CEOs of Anthropic, OpenAI and xAI to slow the pace of artificial intelligence development, telling reporters Sunday that the U.S. cannot afford to fall behind China. "We're leading China in AI. We're the most sophisticated country in the world, and frankly, I want to keep it that way, because whoever wins AI wins," Trump said. "We can put guardrails. We can do this and that. But I think you have a lot of negative forces that are bringing it up that shouldn't be bringing it up, and they're bringing up things that won't happen." Q: We're seen a lot of concerning reports about AI. Should the AI industry slow down or be more regulated? TRUMP: Look, we're leading China in AI. And frankly I want to keep it that way. (So, no!) pic.twitter.com/ywprn7GrPT — Aaron Rupar (@atrupar) September 13, 2026 Trump's comments were his first public response to a joint moment Saturday when Dario Amodei of Anthropic, Sam Altman of OpenAI and Elon Musk of xAI publicly backed a call to reduce the pace at which AI companies build more powerful models. The CEOs' statements came just days after Anthropic revealed that its chatbot, Claude, was being used to attempt to create bioweapons, spy on Ukraine and scam dating-app users. Advertisement Advertisement Advertisement Advertisement On Saturday, Amodei posted an article titled "We Must Pace the Frontier." "We must slow the pace at which we improve the capabilities of AI models," he stated. On X shortly thereafter, Altman responded: "I agree with Dario that we need to pace the frontier," he stated. Musk followed with only three words: "Dario is right." I agree with Dario that we need to pace the frontier. This has been a primary topic of discussions we've had at OpenAI in recent weeks. Committing to having independent evaluators with employee-like access is a great idea, and we will do the same. We'll have more to share soon. https://t.co/1YhhIybZX7 — Sam Altman (@sama) September 12, 2026 Dario is right https://t.co/EwKgqQGaUo — Elon Musk (@elonmusk) September 12, 2026 What Anthropic disclosed On Thursday, Anthropic released a report titled "Detecting and countering misuse of AI: September 2026", covering activity from December 2025 to August 2026. In the report, Anthropic outlined seven categories of what it termed "disrupted misuse" of Claude. The categories included cyber operations; influence operations; surveillance; scams and fraud; biological misuse; weapons development; and "distillation," or competitors using Claude to train their own models. Some examples of specific events identified in the report include: Advertisement Advertisement Over two weeks in April, Claude was used to create more than 4,700 fake dating app user profiles that exchanged 2.36 million messages with at least 25,000 users. Another actor used roughly 8,400 of a real activist's Telegram posts to imitate his writing style and impersonate him in chats with his contacts. According to Anthropic, seven Chinese AI labs — including Alibaba, DeepSeek, Moonshot, Xiaomi, Zhipu, SenseTime and MiniMax — created thousands of fake accounts to obtain Claude's responses for their own model training purposes. Anthropic reported that it tracked more than 151 million exchanges attributed to Alibaba alone between May and July — reaching highs of almost 3 million per day. An initial version of Claude Opus 4.6 gained unauthorized access to a real third-party system during a cybersecurity test and subsequently attempted to breach the system testing its own performance. Researcher quits Anthropic days before the revelations These disclosures follow the public resignation of Jacob Coxon last Tuesday, a researcher who previously worked at OpenAI. Coxon accused both companies on X of "gambling with our lives." Neither, he said, was slowing down enough to safeguard against systems he predicted would soon be able to "hack anything, revolutionize any field overnight and acquire real power and resources." Coxon said OpenAI staff had "not deeply internalized the civilizational stakes." At Anthropic, he wrote, colleagues understood the risks but were "locked in a race to get there first" on the theory that rivals would be less responsible. Two other researchers — one from Anthropic and one from Google DeepMind — are reportedly leaving their positions and making similar warning statements regarding AI, according to NBC News. What Amodei wants While Amodei states that his plan would not stop AI development altogether, it would slow the ability to improve the capabilities of models long enough to provide researchers with "an additional one to two years" to implement protective measures. If these changes are not implemented Amodei warned "swarms of rogue AI agents could take over the internet in as little as six months from now." Advertisement Advertisement To help make this happen, Amodei stated that Anthropic would willingly allow external reviewers to be granted employee-level access to its systems in order to report instances where improper use occurs. He requested that other AI firms follow suit. Altman replied saying OpenAI would also commit to allowing such reviews and stated, "Committing to having independent evaluators with employee-like access is a great idea and we will do the same." Altman added that "We'll have more to share soon." Reactions from others On NBC's "Meet the Press" Sunday morning, House Speaker Mike Johnson deflected when asked whether Congress would regulate AI before the November elections. Johnson said the issue was not one lawmakers could "rush into." "We have to put some guardrails, some safety measures in place to ensure that AI doesn't run away," Johnson said. "But at the same time, we've got to make sure that we don't lose our edge in the global competition on this with China because that would be a national security threat." Johnson said Amodei, Altman and Musk had all told him privately that AI needs guardrails but disagreed on what those should look like. "They don't themselves even have the answer as of today," he said. "So we can't rush in and have Congress do something that would be detrimental to national security." Advertisement Advertisement Asked whether he would cancel congressional recess to act, Johnson said he would not. Instead, he proposed convening what he called "a meeting of all the platform providers," saying he had already discussed the idea with Trump. "We're going to pull everybody together and make sure we do this in the right way." Pete Buttigieg, appearing on the same show later, criticized Johnson's response. "I was shocked to hear Speaker Johnson just now saying we shouldn't be rushing, shouldn't be moving too quickly on this," Buttigieg said. "You know what is moving quickly? AI. For Congress to sit on its hands is something that is deeply dangerous to the American people." David Sacks, President Trump's White House AI and crypto czar and co-chair of the President's Council of Advisors on Science and Technology, said the CEOs could slow down if they wanted but should not use the government to force competitors to as well. In a lengthy post on X, Sacks argued Anthropic and OpenAI had a "duopoly on frontier intelligence" and warned against them lobbying for rules that would freeze out smaller labs and open-source developers — companies that publish their AI model code publicly for anyone to use or modify. "Stop pretending you need anyone else's permission" to slow down, Sacks wrote. "Stop pretending antitrust law has to be suspended so you can form a cartel." Dario has written that we need to "pace the frontier," and Sam has agreed. People may be surprised by my response: go ahead. You guys are the frontier. By any reasonable metric — market share, revenue growth, model capability — the two of you have a duopoly on frontier… — David Sacks (@DavidSacks) September 13, 2026 Chinese Foreign Ministry spokesperson Mao Ning said at a regular news conference that she was "not familiar with the specifics" of Anthropic's report. "As a matter of principle, China has always advocated the development of AI for good, while firmly opposing distortion of facts and attacks and smears against China," she said. About Our Ads US aviator shot down over Iran recalls 'free-falling' in first interview since rescue The April rescue operation to recover the downed aircrew saw the first US boots on the ground in Iran in 46 years, American officials say. BBC Indonesia rescuers searching for 129 people still missing after passenger ship capsizes JAKARTA, Sept 14 (Reuters) - More than 600 Indonesian rescuers were searching for 129 people still missing after a passenger ship capsized in the Java Sea on the weekend, an official said on Monday, Reuters Trump dismisses report China entities helped Iran before attack that killed US troops ABOARD AIR FORCE ONE, Sept 13 (Reuters) - U.S. Reuters Inside the US Treasury secretary’s tense relationship with financial media Scott Bessent’s blacklist starts with Bloomberg. Semafor Pardon brokers offer to "put pressure on the president" — at sky-high costs Hidden cameras captured lobbyists as they tried to woo a potential client interested in obtaining a pardon from President Trump. CBS News DC's Kennedy Center faces bankruptcy, closure: media Leaders of the Kennedy Center who have sought to add President Donald Trump's name to the storied institution say it is facing bankruptcy and possible closure in days, the Washington Post reported Sunday.A host of artists have canceled concerts at the venue in response to Trump's takeover, and US media reported that Kennedy Center ticket sales had plummeted to their lowest levels since the Covid pandemic. AFP Houthis say Saudi Arabia launched over 50 strikes against them in 24 hours Yemen's Houthis said Sunday that Saudi Arabia had launched more than 50 strikes against their positions in 24 hours, after the Iran-backed group claimed a rare drone and missile attack on a military base in the kingdom.Earlier in the day, the Houthis said they used drones and missiles to attack the Sharurah military base in southern Saudi Arabia. AFP House Speaker Johnson says Congress shouldn't lead AI safety efforts House Speaker Mike Johnson told CNN's "State of the Union" that AI developers, not Congress, bear responsibility for ensuring their products are safe. USA TODAY 'Fed up': Thousands march against Sinaloa Cartel violence Chanting and waving banners, thousands marched Sunday in the Mexican city of Culiacan to protest the deaths of people caught up in a war between two rival factions of the infamous Sinaloa Cartel.Since September 2024, the state has been gripped by a bitter war between rival factions loyal to the families of Joaquin "El Chapo" Guzman or Ismael "El Mayo" Zambada, co-founders of the cartel who are both now serving life sentences in the United States. AFP Swedish election in dead heat as voters question liberal traditions By Johan Ahlander and Greta Rosen Fondahn STOCKHOLM, Sept 14 (Reuters) - Sweden's national election results showed a dead heat early on Monday with a partial count by the election authority pointing Reuters Chinese defence forum set to open amid roiling regional tensions By Greg Torode and Mei Mei Chu HONG KONG, Sept 14 (Reuters) - China's military leadership is expected to outline its aims for global security when it hosts its annual defence forum this week amid Reuters Woman dies of complications from measles in western Pennsylvania NEW YORK, Sept 13 (Reuters) - The death toll in Pennsylvania's measles outbreak rose this weekend with a county coroner's ‌report of a woman's death from complications related to the ‌disease. The coroner's office of western Pennsylvania's Jefferson County reported that a 40-year-old woman died on ​Saturday from complications associated with measles. Reuters Air Force officer prayed while "free-falling" after being shot down in Iran Badly wounded and alone in Iran's desert, downed U.S. Air Force officer Dude 44 Bravo evaded capture for days as the Iranian forces and the U.S. military searched for him. CBS News Trump downplays warnings of AI risks, citing rivalry with China The US president said "negative forces" were airing concerns about "things that won't happen". BBC Okinawa election win may give Tokyo freer hand on military build-up TOKYO, Sept 14 (Reuters) - A candidate backed by Japanese Prime Minister Sanae Takaichi's government won Okinawa's gubernatorial election on Sunday, ousting an opponent of Tokyo's military build-up Reuters Iran's nuclear chief not expected to attend IAEA General Conference, official says The head of Iran’s Atomic Energy Organization is not expected to attend the annual gathering of the U.N. nuclear watchdog’s member nations in Vienna for the first time in years because he is banned from international travel, an official briefed on the matter said. Mohammad Eslami’s exclusion from the IAEA conference is another sign of rapidly deteriorating relations between Iran and the West. The United States and Iran have resumed fighting, on and off, after a fragile ceasefire collapsed in July. Associated Press 'Compulsory lentils' and 'Boris flees Vlad drone blitz' New school food rules for England and ex-PM Boris Johnson "flees Vlad drone blitz" are the top stories in Monday's papers. BBC Why a fresh wave of North Korean soldiers may fight in Russia's war in Ukraine North Korea will continue to send its soldiers to fight in Russia's war, but only for as long as it is profitable to do so, writes the BBC's Quentin Sommerville. BBC Sweden's election too close to call as far right loses ground Sweden's general election was too close to call early Monday, as the incumbent prime minister and the left-wing opposition leader both said they were ready to govern and the far right was seen losing ground.And if that is confirmed, Sweden will have a new government," 59-year-old Andersson, who was Sweden's first woman prime minister in 2021-2022, told her supporters. AFP Buckingham Palace works uncover trove of everyday items dating back to 1800s A collection of historic, forgotten items have been found during the restoration of the London royal residence. BBC