Who is behind the "AI audit" that accuses Emmanuel Saez and Gabriel Zucman of "errors"?
An anonymous audit claiming to have identified over 2,000 errors in the research of economists Emmanuel Saez and Gabriel Zucman was published on October 28, 2026, via the blog of Tyler Cowen. The audit utilized artificial intelligence models, specifically Claude Opus 5.5 and GPT-6 Astra, to analyze 138 publications by the researchers and their co-authors. While initial claims suggested 2,035 issues, an updated version revised these figures to 53 uncontested defects and 94 debatable methodological choices. Saez and Zucman condemned the exercise as a grotesque misuse of AI intended to discredit their work ahead of California’s Proposition 40 vote on a billionaire tax scheduled for November 3, 2026. Many economists, including critics of Saez and Zucman, dismissed the audit as unconvincing and unfair, noting that the findings had not been confirmed by independent scientific expertise.
Californian voters rejected Proposal 40, a proposed one-time 5 percent tax on the wealth of billionaires, during the midterm elections held on November 3, 2026. Preliminary results indicated that more than 55 percent of voters opposed the measure, while fewer than 45 percent supported it, based on approximately 60 percent of ballots counted. The defeat occurred despite the proposal being championed by prominent left-wing figures and economists who argued for increased taxation of extreme wealth. The initiative, which aimed to levy a flat fee on individuals with a net worth exceeding one billion dollars, was intended to generate revenue primarily to offset federal cuts in healthcare funding. Although the specific tax failed at the ballot box, the intense campaign surrounding it sparked significant controversy within the economic community and among political leaders. The measure, known as Proposal 40, was designed to impose a temporary tax on residents of California whose net assets surpassed one billion dollars as of January 1, 2026. The full five percent rate applied to wealth above 1.1 billion dollars. Proponents estimated that the tax would yield approximately 100 billion dollars between 2027 and 2031, allowing taxpayers to pay in installments over five years. Ninety percent of the projected revenue was earmarked for healthcare, with the remainder allocated to education, food assistance, and administrative costs. The Legislative Analyst’s Office, an independent body providing economic analysis to the California legislature, described the potential revenue as amounting to tens of billions of dollars. The proposal was promoted by the union SEIU-United Healthcare Workers West and received backing from Senator Bernie Sanders, who called it the most important referendum in the country. Nobel laureates Joseph Stiglitz, Paul Krugman, Daron Acemoglu, Peter Diamond, Abhijit Banerjee, and Esther Duflo signed an open letter supporting the measure, noting that the wealth of Californian billionaires had tripled over the last decade to total 2.3 trillion dollars, while their income taxes accounted for only 1.6 percent of their asset growth between 2019 and 2025. Two additional ballot measures, Proposals 41 and 42, were included on the same ballot with the explicit goal of weakening or blocking Proposal 40. These initiatives sought to prevent the application of the billionaire tax even if Proposal 40 passed. Opponents of the tax, led by wealthy technology entrepreneurs, warned that the levy would trigger an exodus of high-net-worth individuals to other states, potentially reducing overall state revenue. The debate extended beyond California, influencing discussions in France regarding similar fiscal policies inspired by economist Thomas Piketty. In the final days before the vote, a dispute erupted over the academic integrity of the researchers behind the proposal. On October 28, Tyler Cowen, an economics professor at George Mason University and host of the popular blog Marginal Revolution, published a post relaying findings from an anonymous audit. The audit claimed to have identified errors in the work of Emmanuel Saez and Gabriel Zucman, two leading inequality scholars and key architects of the tax idea. In its initial release, the audit alleged that 2,035 problems existed across 138 publications by Saez, Zucman, and their co-authors, with 147 issues deemed severe enough to alter major results. It suggested that 69 percent of the papers contained at least one error and that 97 percent of significant or moderate errors skewed conclusions toward overestimation. The methodology behind this audit drew sharp criticism. According to the report, two artificial intelligence models, Claude Opus 5.5 and GPT-6 Astra, independently examined the publications, including appendices and replication files. A third AI model subsequently reviewed the flagged issues, attempted to refute them, and categorized their severity. The AI systems were provided with a corpus of 131 previously published criticisms by other economists to test known objections. However, the audit did not constitute an independent scientific review following standard peer-review procedures. Furthermore, the AI models lacked access to confidential data available only to the original researchers. By October 29, an updated version of the audit revised its claims, distinguishing 53 indisputable defects from 94 methodological choices that were debatable but defensible. The identity of the auditor remained unknown, with no public information regarding their academic background or statistical expertise. Many economists criticized the validity of the AI-driven review. Critics labeled the exercise as nonsensical and unjust, arguing it lacked scientific rigor. Saez and Zucman condemned the publication as a grotesque misuse of artificial intelligence. They viewed the timing, just six days before the election, as an attempt to discredit their research. Zucman stated that the audit held no value or interest. Despite these defenses, the controversy gained massive traction online. Cowen’s post garnered hundreds of thousands of views within hours, reaching 1.2 million views by October 31. Tech magnate Elon Musk commented on the story with the phrase "no kidding." The broader campaign against the tax featured heated exchanges through op-eds, television debates, and social media posts. Liberal essayist Rainer Zitelmann argued that although the tax failed in California, a stronghold of American liberalism, the prolonged campaign by French economists had already caused damage. He suggested that the rhetoric employed by figures associated with the policy shifted toward populism. Economists Thesmar and Landier noted that such a tax would not resolve the structural challenges facing the welfare state. The failure of Proposal 40 marked a setback for advocates of wealth taxation in the United States, though the underlying debate over inequality and fiscal policy continues to influence political discourse in both America and Europe.
Go to the primary sources (1)
The official sources this coverage is built on. Read them directly to bypass framing.
On November 3, 2026, California voters will decide on a citizen initiative proposing a one-time 5% tax on the wealth of billionaires. The measure was co-authored by French economist Emmanuel Saez, who is based at the University of Berkeley alongside Gabriel Zucman, another prominent figure in discussions regarding wealth taxation. Polls indicate that the election result is expected to be very close. This local ballot measure is part of the broader U.S. midterm elections occurring in 2026. The article highlights the academic origins of the proposal within the Berkeley economics department.
Bias read (Center): The article presents a factual report on a specific ballot measure and its academic origins. It cites polling data indicating a close race and identifies the authors of the proposal without using loaded language to praise or condemn the tax itself. While the subject matter involves progressive tax政策
Why factuality: no official source document/info detected
Why objectivity (90): The tone is journalistic and descriptive rather than argumentative. While the phrase 'measure that makes great fortunes tremble' adds a touch of dramatic flair typical of feature reporting, it does not constitute heavy bias. The piece focuses on the human element (Saez's background) and the setting,
The article reports that California voters rejected Proposition 40, a wealth tax targeting billionaires supported by economists Gabriel Zucman and Emmanuel Saez. Preliminary results from the November 3 referendum indicate that over 55% of voters opposed the measure while less than 45% supported it, despite the state being considered a stronghold of American left-wing politics. The piece features commentary from liberal essayist Rainer Zitelmann, who argues that although the specific tax failed, the prolonged campaign by French economists has already caused negative effects. This analysis appears in the 'Ideas and Debates' section of L'Express, highlighting the intersection of international economic theory and US local policy.
Bias read (Conservative): The article frames the rejection of a progressive wealth tax as a victory against harmful leftist policies, relying on a conservative commentator to characterize the campaign's impact negatively rather than presenting balanced views on the tax itself.
The article examines the upcoming California ballot measure, known as Proposition 40, which proposes a temporary 5% wealth tax on billionaires. It highlights the intense debate among economists regarding this initiative, noting that it was inspired by French scholars Gabriel Zucman and Emmanuel Saez. The piece features an interview with professors Thesmar and Landier, who argue that such a 'Zucman tax' would fail to solve fundamental issues within the welfare state. The authors further suggest that proponents like Piketty and Zucman have shifted toward populist rhetoric in their advocacy for these fiscal policies.
Bias read (Conservative): The headline explicitly labels prominent economists as adopting 'populist rhetoric,' while the subhead asserts the proposed tax will not solve state problems, reflecting a critical stance against progressive fiscal policies.
Voters in California are set to decide on Proposition 40, a ballot measure proposing a one-time 5% tax on the net worth of residents exceeding $1 billion. The initiative aims to generate revenue primarily to offset recent federal cuts to healthcare funding. This proposal marks a potential historic shift, as California would become the first US state to implement a wealth tax targeting billionaires. The campaign surrounding the measure has been described as one of the most expensive in state history, reflecting intense opposition from the ultra-wealthy who are considering relocation to avoid the levy. Additionally, the proposal has caused significant division among national liberal political leaders regarding its strategic implications.
Bias read (Center): The text uses neutral, descriptive language such as 'divided' and 'scramble,' presenting the economic stakes and political conflict without adopting a partisan stance or using loaded adjectives.
Voters in California will decide on October 3, during the midterm elections, whether to approve a novel tax measure. The proposal imposes a 5% levy on the wealth of billionaires to fund public initiatives, including healthcare services. This potential fiscal change has triggered strong opposition from wealthy individuals within the Silicon Valley technology sector.
Bias read (Conservative): The headline uses emotionally charged language like 'makes tech billionaires go crazy' and frames the story around the anger of the rich rather than the policy merits.
An anonymous audit claiming to have identified over 2,000 errors in the research of economists Emmanuel Saez and Gabriel Zucman was published on October 28, 2026, via the blog of Tyler Cowen. The audit utilized artificial intelligence models, specifically Claude Opus 5.5 and GPT-6 Astra, to analyze 138 publications by the researchers and their co-authors. While initial claims suggested 2,035 issues, an updated version revised these figures to 53 uncontested defects and 94 debatable methodological choices. Saez and Zucman condemned the exercise as a grotesque misuse of AI intended to discredit their work ahead of California’s Proposition 40 vote on a billionaire tax scheduled for November 3, 2026. Many economists, including critics of Saez and Zucman, dismissed the audit as unconvincing and unfair, noting that the findings had not been confirmed by independent scientific expertise.
Bias read (Progressive): The article frames the AI audit as a politically motivated smear campaign targeting proponents of progressive taxation just before a key election, emphasizing the lack of independent verification and the timing relative to the ballot measure.
The article examines the debate surrounding two proposed wealth taxes targeting billionaires: a 5% one-time tax on assets in California (Proposition 40) and a national solidarity tax concept in France. It notes that these proposals are heavily influenced by the academic work of economists Gabriel Zucman and Emmanuel Saez in the US, and Thomas Piketty in France. The piece highlights the intensity of the public discourse, which has been characterized by heated exchanges through op-eds, televised debates, and extensive discussions on social media platforms like X. This context sets the stage for analyzing the political and economic implications of taxing high-net-worth individuals.
Bias read (Progressive): The headline uses the pejorative term 'surenchère' (overbidding/exaggeration) to frame the policies, while the body text neutrally describes the intense public debate and cites prominent left-wing economists as the intellectual inspiration for the measures.
An AI-generated audit, shared by prominent economist Tyler Cowen, claims that research by Gabriel Zucman and Emmanuel Saez contains hundreds of significant errors across 138 papers. The report identifies over 2,000 issues, including 147 deemed severe enough to alter major findings regarding economic inequality. This revelation has sparked intense debate within the academic community regarding the reliability of automated auditing methods. The timing coincides closely with a California referendum on a billionaire tax, which Saez helped design, adding political urgency to the scientific dispute. High-profile figures, including Elon Musk, have commented on the controversy, amplifying its reach beyond traditional academic circles.
Bias read (Center): The article presents the controversy neutrally, citing both the specific allegations and the ongoing academic debate about the audit's validity without taking a side.
How each side covered it
The same event, grouped by the political lean of the outlets covering it.
progressive
center
conservative
★
How each side covered it
Support independent, bias-aware news and unlock the social pulse, community voting, and every other Supporter feature.
France
