Editor’s note, May 18, 2024, 7:30 pm ET: This story has been updated to reflect OpenAI CEO Sam Altman’s tweet on Saturday afternoon that the company was in the process of changing its offboarding documents. For months, OpenAI has been losing employees who care deeply about making sure AI is safe. Now, the company is positively hemorrhaging them. Ilya Sutskever and Jan Leike announced their departures from OpenAI, the maker of ChatGPT, on Tuesday. They were the leaders of the company’s superalignment team — the team tasked with ensuring that AI stays aligned with the goals of its makers, rather than acting unpredictably and harming humanity. They’re not the only ones who’ve left. Since last November — when OpenAI’s board tried to fire CEO Sam Altman only to see him quickly claw his way back to power — at least five more of the company’s most safety-conscious employees have either quit or been pushed out. What’s going on here? If you’ve been following the saga on social media, you might think OpenAI secretly made a huge technological breakthrough. The meme “What did Ilya see?” speculates that Sutskever, the former chief scientist, left because he saw something horrifying, like an AI system that could destroy humanity. But the real answer may have less to do with pessimism about technology and more to do with pessimism about humans — and one human in particular: Altman. According to sources familiar with the company, safety-minded employees have lost faith in him. “It’s a process of trust collapsing bit by bit, like dominoes falling one by one,” a person with inside knowledge of the company told me, speaking on condition of anonymity. Not many employees are willing to speak about this publicly. That’s partly because OpenAI is known for getting its workers to sign offboarding agreements with non-disparagement provisions upon leaving. If you refuse to sign one, you give up your equity in the company, which means you potentially lose out on millions of dollars. (OpenAI did not respond to a request for comment in time for publication. After publication of my colleague Kelsey Piper’s piece on OpenAI’s post-employment agreements, OpenAI sent her a statement noting, “We have never canceled any current or former employee’s vested equity nor will we if people do not sign a release or nondisparagement agreement when they exit.” When Piper asked if this represented a change in policy, as sources close to the company had indicated to her, OpenAI replied: “This statement reflects reality.” On Saturday afternoon, a little more than a day after this article published, Altman acknowledged in a tweet that there had been a provision in the company’s off-boarding documents about “potential equity cancellation” for departing employees, but said the company was in the process of changing that language.) [Image: https://platform.vox.com/wp-content/uploads/sites/2/chorus/uploads/chorus_asset/file/25454608/Screenshot_2024_05_18_at_6.33.22_PM.png?quality=90&strip=all] One former employee, however, refused to sign the offboarding agreement so that he would be free to criticize the company. Daniel Kokotajlo, who joined OpenAI in 2022 with hopes of steering it toward safe deployment of AI, worked on the governance team — until he quit last month. “OpenAI is training ever-more-powerful AI systems with the goal of eventually surpassing human intelligence across the board. This could be the best thing that has ever happened to humanity, but it could also be the worst if we don’t proceed with care,” Kokotajlo told me this week. OpenAI says it wants to build artificial general intelligence (AGI), a hypothetical system that can perform at human or superhuman levels across many domains. “I joined with substantial hope that OpenAI would rise to the occasion and behave more responsibly as they got closer to achieving AGI. It slowly became clear to many of us that this would not happen,” Kokotajlo told me. “I gradually lost trust in OpenAI leadership and their ability to responsibly handle AGI, so I quit.” And Leike, explaining in a thread on X why he quit as co-leader of the superalignment team, painted a very similar picture Friday. “I have been disagreeing with OpenAI leadership about the company’s core priorities for quite some time, until we finally reached a breaking point,” he wrote. Why OpenAI’s safety team grew to distrust Sam Altman To get a handle on what happened, we need to rewind to last November. That’s when Sutskever, working together with the OpenAI board, tried to fire Altman. The board said Altman was “not consistently candid in his communications.” Translation: We don’t trust him. The ouster failed spectacularly. Altman and his ally, company president Greg Brockman, threatened to take OpenAI’s top talent to Microsoft — effectively destroying OpenAI — unless Altman was reinstated. Faced with that threat, the board gave in. Altman came back more powerful than ever, with new, more supportive board members and a freer hand to run the company. When you shoot at the king and miss, things tend to get awkward. Publicly, Sutskever and Altman gave the appearance of a continuing friendship. And when Sutskever announced his departure this week, he said he was heading off to pursue “a project that is very personally meaningful to me.” Altman posted on X two minutes later, saying that “this is very sad to me; Ilya is … a dear friend.” Yet Sutskever has not been seen at the OpenAI office in about six months — ever since the attempted coup. He has been remotely co-leading the superalignment team, tasked with making sure a future AGI would be aligned with the goals of humanity rather than going rogue. It’s a nice enough ambition, but one that’s divorced from the daily operations of the company, which has been racing to commercialize products under Altman’s leadership. And then there was this tweet, posted shortly after Altman’s reinstatement and quickly deleted: [Image: https://platform.vox.com/wp-content/uploads/sites/2/chorus/uploads/chorus_asset/file/25451798/Screenshot_2024_05_16_at_5.05.54_PM.png?quality=90&strip=all] So, despite the public-facing camaraderie, there’s reason to be skeptical that Sutskever and Altman were friends after the former attempted to oust the latter. And Altman’s reaction to being fired had revealed something about his character: His threat to hollow out OpenAI unless the board rehired him, and his insistence on stacking the board with new members skewed in his favor, showed a determination to hold onto power and avoid future checks on it. Former colleagues and employees came forward to describe him as a manipulator who speaks out of both sides of his mouth — someone who claims, for instance, that he wants to prioritize safety, but contradicts that in his behaviors. For example, Altman was fundraising with autocratic regimes like Saudi Arabia so he could spin up a new AI chip-making company, which would give him a huge supply of the coveted resources needed to build cutting-edge AI. That was alarming to safety-minded employees. If Altman truly cared about building and deploying AI in the safest way possible, why did he seem to be in a mad dash to accumulate as many chips as possible, which would only accelerate the technology? For that matter, why was he taking the safety risk of working with regimes that might use AI to supercharge digital surveillance or human rights abuses? For employees, all this led to a gradual “loss of belief that when OpenAI says it’s going to do something or says that it values something, that that is actually true,” a source with inside knowledge of the company told me. That gradual process crescendoed this week. The superalignment team’s co-leader, Jan Leike, did not bother to play nice. “I resigned,” he posted on X, mere hours after Sutskever announced his departure. No warm goodbyes. No vote of confidence in the company’s leadership. Other safety-minded former employees quote-tweeted Leike’s blunt resignation, appending heart emojis. One of them was Leopold Aschenbrenner, a Sutskever ally and superalignment team member who was fired from OpenAI last month. Media reports noted that he and Pavel Izmailov, another researcher on the same team, were allegedly fired for leaking information. But OpenAI has offered no evidence of a leak. And given the strict confidentiality agreement everyone signs when they first join OpenAI, it would be easy for Altman — a deeply networked Silicon Valley veteran who is an expert at working the press — to portray sharing even the most innocuous of information as “leaking,” if he was keen to get rid of Sutskever’s allies. The same month that Aschenbrenner and Izmailov were forced out, another safety researcher, Cullen O’Keefe, also departed the company. And two weeks ago, yet another safety researcher, William Saunders, wrote a cryptic post on the EA Forum, an online gathering place for members of the effective altruism movement, who have been heavily involved in the cause of AI safety. Saunders summarized the work he’s done at OpenAI as part of the superalignment team. Then he wrote: “I resigned from OpenAI on February 15, 2024.” A commenter asked the obvious question: Why was Saunders posting this? “No comment,” Saunders replied. Commenters concluded that he is probably bound by a non-disparagement agreement. Putting all of this together with my conversations with company insiders, what we get is a picture of at least seven people who tried to push OpenAI to greater safety from within, but ultimately lost so much faith in its charismatic leader that their position became untenable. “I think a lot of people in the company who take safety and social impact seriously think of it as an open question: is working for a company like OpenAI a good thing to do?” said the person with inside knowledge of the company. “And the answer is only ‘yes’ to the extent that OpenAI is really going to be thoughtful and responsible about what it’s doing.” With the safety team gutted, who will make sure OpenAI’s work is safe? With Leike no longer there to run the superalignment team, OpenAI has replaced him with company co-founder John Schulman. But the team has been hollowed out. And Schulman already has his hands full with his preexisting full-time job ensuring the safety of OpenAI’s current products. How much serious, forward-looking safety work can we hope for at OpenAI going forward? Probably not much. “The whole point of setting up the superalignment team was that there’s actually different kinds of safety issues that arise if the company is successful in building AGI,” the person with inside knowledge told me. “So, this was a dedicated investment in that future.” Even when the team was functioning at full capacity, that “dedicated investment” was home to a tiny fraction of OpenAI’s researchers and was promised only 20 percent of its computing power — perhaps the most important resource at an AI company. Now, that computing power may be siphoned off to other OpenAI teams, and it’s unclear if there’ll be much focus on avoiding catastrophic risk from future AI models. To be clear, this does not mean the products OpenAI is releasing now — like the new version of ChatGPT, dubbed GPT-4o, which can have a natural-sounding dialogue with users — are going to destroy humanity. But what’s coming down the pike? “It’s important to distinguish between ‘Are they currently building and deploying AI systems that are unsafe?’ versus ‘Are they on track to build and deploy AGI or superintelligence safely?’” the source with inside knowledge said. “I think the answer to the second question is no.” Leike expressed that same concern in his Friday thread on X. He noted that his team had been struggling to get enough computing power to do its work and generally “sailing against the wind.” [Image: https://platform.vox.com/wp-content/uploads/sites/2/chorus/uploads/chorus_asset/file/25452835/Screenshot_2024_05_17_at_12.11.21_PM.png?quality=90&strip=all] Most strikingly, Leike said, “I believe much more of our bandwidth should be spent getting ready for the next generations of models, on security, monitoring, preparedness, safety, adversarial robustness, (super)alignment, confidentiality, societal impact, and related topics. These problems are quite hard to get right, and I am concerned we aren’t on a trajectory to get there.” When one of the world’s leading minds in AI safety says the world’s leading AI company isn’t on the right trajectory, we all have reason to be concerned. Update, May 18, 7:30 pm ET: This story was published on May 17 and has been updated multiple times, most recently to include Sam Altman’s response on social media.

Americans have been pessimistic about the economy for years. Weirdly, that’s seemed to have little impact on their willingness to open their wallets. Retail sales surged during the pandemic as home-bound workers clicked “complete purchase” on everything from Pelotons to sourdough starter. In 2020, e-commerce sales rose by 43 percent. Stimulus checks gave Americans newfound savings and excess money to burn. Supply chains couldn’t keep up with the demand. That was all supposed to come crashing down at some point. For more than a year, economists warned about the “death of the consumer” and a resulting recession — neither of which have materialized. Consumers were expected to retreat as inflation skyrocketed, hitting a 9.1 percent peak in June 2022 and remaining stubbornly above the Federal Reserve’s target rate of 2 percent. Instead, Americans just kept buying more, even accounting for price increases and beyond growth in their disposable income. Their spending helped drive US economic growth in 2023 and remained high in the first months of this year. In March, consumer spending increased by 0.8 percent, exceeding expectations from financial analysts. [Image: https://platform.vox.com/wp-content/uploads/sites/2/chorus/uploads/chorus_asset/file/25453285/uqyyM_what_americans_are_buying_in_2024__5_.png?quality=90&strip=all] There is a sign that Americans’ shopping spree might be finally coming to an end: Retail spending stayed the same in April as compared to the previous month, falling short of analyst projections for growth. However, those numbers don’t capture spending on services — for example, health care, transportation, and insurance — which has increased markedly this year. And both Preston Caldwell, a senior US economist at Morningstar, and Scott Hoyt, a Moody’s Analytics economist, said those numbers could easily bounce back next month, even if they’re expecting spending to cool by the end of the year. “I am anticipating that we do see eventually a consumer slowdown over the course of this year,” Caldwell said. “It’s premature to say that that’s already playing out right now.” Indeed, spending is bound to come down at some point under the pressure of high interest rates, which the Fed isn’t expected to cut until later this year — or potentially at all in 2024. So why, despite all the doom and gloom among consumers, has spending been so resilient? Who is driving high spending? Two things are simultaneously true: People feel really negatively about the economy, and that’s not stopping them from spending. In May, the University of Michigan recorded its lowest consumer sentiment reading in six months — an index of 67.4 out of 100 — as part of its long-running survey. That’s up from this time last year, but still well below pre-pandemic consumer sentiment readings, which hovered in the upper 80s and 90s. The trend in sentiment was widespread across demographic lines: Consumers “expressed worries that inflation, unemployment and interest rates may all be moving in an unfavorable direction in the year ahead,” the University of Michigan report reads. It’s hard to reconcile that with high spending figures. But in short, the rich currently feel rich and account for a large share of overall spending. The middle class feels a little better off too, and likely still has some savings built up they can burn through. They might not yet have felt the pressure of high interest rates and inflation to the same degree as people who rent and have fewer investments. (But that’s due to change.) High-income consumers — households in the top 20 percent of income earning at least $244,025 before taxes as of 2022 — have been largely cushioned from economic headwinds and are flush with cash to spend. The pandemic saw Americans’ average percentage of income saved increase to an all-time high of 32 percent in April 2020 after many households received stimulus checks. That has helped fuel spending, but unlike in other high-income countries where consumers have proved more thrifty, Americans are close to depleting those savings. “The excess savings [are] still kind of sloshing their way through the system. Depending on how you estimate excess savings, they will be depleted sometime in the middle of 2024 or maybe by as late as mid-2025,” Caldwell said. Many high-income consumers also locked in low interest rates on their mortgages before the Federal Reserve started raising rates in March 2022, and they’re seeing their home values continue to go up nonetheless. The average US home price increased from $287,000 in 2019 to $450,000 in 2024. This is in part due to persistent low inventory: High interest rates have kept would-be sellers on the sidelines because their mortgage payments would be higher if they bought a new place. High-income consumers have also seen their investment portfolios balloon in the last year. The stock market repeatedly tested new highs in recent months, with the latest record set on Thursday in the wake of new data showing that inflation is slowing. And wealthy older Americans who allocate more of their portfolios to government bonds are benefiting from higher interest rates. “That sort of gives consumers an incentive to spend out of their newfound wealth,” Hoyt said. “And since this set of consumers still has excess savings left over from the pandemic, that gives them the easy, relatively liquid monies to do so.” The question is how long the stock market can sustain this run. Some analysts think stocks are currently overpriced and due for a correction — which might cause some people to finally close their wallets. “Equity prices are starting to move more into arguably overvalued territory,” Caldwell said. “So that’s probably not going to be a tailwind [for spending] over the next year.” At the same time, the factors currently fueling spending at the highest income levels aren’t universal. Not all consumers can afford to spend more. Even though inflation has come down significantly from its 2022 peak, low-income Americans are struggling with higher prices. Consumers in general say they are budgeting more on everyday essentials like fresh produce and baby supplies. Among people living paycheck to paycheck, pandemic savings (if they ever really had any) might be long gone. Low and moderate-income consumers are also increasingly weighed down by credit card debt and struggling to pay it down due to high interest rates, which research suggests could be a major contributor to overall economic pessimism. Though credit card debt levels dipped during the pandemic, they are now returning to pre-pandemic levels, with the average balance per consumer increasing by 8.5 percent in the last year to $6,218. More than half of people earning less than $25,000 carry a balance on their credit cards. [Image: https://platform.vox.com/wp-content/uploads/sites/2/chorus/uploads/chorus_asset/file/25453287/o1V8L_americans_are_racking_up_record_credit_card_debt__2_.png?quality=90&strip=all] Their only consolation is that the job market remains strong, meaning they might be able to count on another paycheck — but even that might not last. Analysts including Caldwell expect the unemployment rate to rise from 3.8 percent to 3.9 percent and wage growth to slow in 2024. Ultimately, however, low-income consumers “just don’t account for that big of a share of total spending,” Hoyt said. “It’s the high end of the income distribution that accounts for a disproportionate share of the spending.”

Donald Trump advertises his authoritarianism like it’s a golf course adorned with his name. The presumptive GOP nominee has repeatedly promised to sic the Justice Department on his political adversaries, vowing to appoint “a real special prosecutor to go after” President Joe Biden, “the entire Biden crime family, and all others involved with the destruction of our elections, borders and our country itself.” He has repeatedly praised the extrajudicial killing of looters and drug dealers, and implored police officers to brutalize criminal suspects. But Trump’s attitude toward lawbreakers who are aligned with his movement is decidedly more lenient. He has repeatedly assured those who commit violence on his behalf — like the January 6 rioters who tried to forestall the peaceful transfer of power in 2021 — that he will immunize them from legal accountability through presidential pardons. Thus, the frontrunner in America’s 2024 election has adopted a gangster’s mentality toward crime: the criminality of any given action is determined by its compatibility with his interests, not the law. In theory, the constitution — with its elaborate division of powers — should constrain Trump’s assaults on the rule of law. That’s surely true to a point. But if Trump’s authoritarian impulses are backed by his fellow Republicans, then the structural constraints on his power in a second term would be less than reliable. Unfortunately, two recent developments indicate that the long arc of Republican politics is bending toward lawlessness. Texas just let a far-right radical get away with murder First, in Texas, you can commit murder without suffering the legal consequences of that crime, so long as your victim’s politics are loathed by the right and your case is championed by conservative media. Or at least, this is the message sent by Gov. Greg Abbott’s pardoning of Daniel Perry. In the weeks after George Floyd’s murder in 2020, the proliferation of Black Lives Matter protests had filled Perry with apparent bloodlust. Then an active-duty Army officer, Perry texted and messaged friends, among other things: * “I might go to Dallas to shoot looters.” * “I might have to kill a few people on my way to work they are rioting outside my apartment complex … No protesters go near me or my car.” * “I wonder if they will let [me] cut the ears off of people who’s decided to commit suicide by me.” When a friend of Perry asked him if he could “catch me a negro daddy,” Perry replied, “That is what I am hoping.” Weeks later, Perry was driving an Uber in Austin, Texas, when he came upon a Black Lives Matter march. According to prosecutors, Perry ran a red light and drove his vehicle into the crowd, almost hitting several protesters. Activists gathered angrily around Perry’s car. Garrett Foster, a 28-year-old Air Force veteran who was openly carrying an AK-47 rifle, approached Perry’s window. Perry then shot Foster dead. At trial, Perry’s defense team alleged that Foster had pointed his rifle at the defendant. But witnesses testified that Foster never brandished his weapon, only carried it, which is legal in Texas. And Perry corroborated that account in his initial statement to the police, saying, “I believe he was going to aim at me. I didn’t want to give him a chance to aim at me.” A jury convicted Perry of murder last year. But this week, the governor of Texas used his pardoning power to release Perry from prison. In a statement, Abbott said, “Texas has one of the strongest ‘stand your ground’ laws of self-defense that cannot be nullified by a jury or a progressive district attorney.” He noted that in the Lone Star State, a person is justified in using deadly force against another if they “reasonably believe the deadly force is immediately necessary” for averting one’s own violent death. The Texas governor argued that it was reasonable for Perry to believe his life was at stake since Foster had held his gun in the “low-ready firing position.” Yet this claim is inconsistent with Perry’s own remarks to the police, which indicated that Foster did not aim a rifle at his killer, but merely carried it. Needless to say, seeing a person lawfully carrying a firearm cannot give one a legal right to kill them. But pesky realities like this carry less weight than conservative media’s delusional grievances. Shortly after Perry’s conviction in April 2023, then-Fox News anchor Tucker Carlson aired a segment portraying Perry as a helpless victim of “a mob of rioters” and a “Soros-funded” district attorney. Carlson decried the jury’s verdict as a “legal atrocity” and lambasted Abbott for standing idly by while his state invalidated conservatives’ right to defend themselves. “So that is Greg Abbott’s position,” he said. “There is no right of self-defense in Texas.” The next day, Abbott pledged to work “as swiftly as Texas law allows regarding the pardon of Sgt. Perry.” Republicans are making it clear they can’t be trusted to check Trump’s most lawless impulses During a second Trump presidency, the independent power of Democratic officials might limit the reach of his authoritarian machinations. A Democratic House or Senate would serve as a check on illiberal legislation, while blue states could leverage their own constitutional authority to impede legally dubious executive orders. But as Abbott’s conduct shows, we should not trust Republican politicians to defend the rule of law. Like Trump, many in the conservative movement believe that its supporters should be held to a more lenient legal standard than its enemies. And they also evince some sympathy for political violence aimed at abetting right-wing power. Crucially, this illiberal faction of the GOP seems to include some Supreme Court justices. To this point, the Roberts Court has checked some of Trump’s more egregious affronts to the constitutional order. Should the GOP secure the opportunity to build an even larger conservative majority, however, that could change. This week, Americans received a reminder of just how radical the Supreme Court’s most right-wing justices have become. In the weeks following the January 6 insurrection, die-hard Trump supporters across the country hung upside-down flags in protest of Biden’s supposed theft of the election. On Thursday, the New York Times reported that one such flag had hung outside the home of Justice Samuel Alito, even as he was presiding over judicial challenges to the 2020 election’s results. Alito claims he had no involvement in the flying of the flag, which his wife had hung upside down in response to “a neighbor’s use of objectionable and personally insulting language on yard signs.” Notably, this explanation does not deny the political meaning of that symbol in January 2021. Alito joined Justices Clarence Thomas and Neil Gorsuch in dissenting from the court’s decision not to hear a challenge to election procedures in Pennsylvania. Thomas’s wife, the conservative activist Ginni Thomas, had also publicly signaled support for the January 6 demonstrators. If Trump secures the opportunity to appoint additional Supreme Court justices, it is all but certain that they will be at least as sympathetic to his extremism as Alito or the Thomas family. None of this means that Trump’s election would mark the end of the American republic. But it does suggest that both Trump and the conservative movement arrayed behind him pose an intolerable threat to the most liberal and democratic features of our system of government.