All-In #291: most of the figures check out. The framing often doesn't.
We checked 14 claims from the October 2 episode against federal data and the outlets the hosts quoted. Seven hold up, four need context, and three don't. Two of those three are large figures: the share of U.S. wealth the middle class holds, and inflation under Trump.
The superintelligence accord was signed by "the six major frontier model companies"
“We worked out this agreement that has come to be known as the White House Accord on Super Intelligence that was then signed by the six major frontier model companies.”
What the record shows
Six companies signed alongside President Trump on Sept 29, 2026: Google, Anthropic, Meta, OpenAI, xAI and NVIDIA. NVIDIA is mainly a chipmaker, so calling all six "frontier model companies" is loose. OpenAI signed through its president, Greg Brockman.
“Dario from Anthropic is obviously very anti-Trump. He has never shown up for these things. The organization is 96% anti-Trump, left-leaning, etc.”
What the record shows
We found no source for a 96% figure. The numbers in circulation measure political donations, not staff views. The most shared one, a post on X by Katie Miller, says more than $200 million in donations tied to Anthropic since 2020 went 99.8% to Democrats. That figure counts people and investors linked to the company. About 59 in every 1,000 Anthropic employees gave to a federal campaign this cycle, so donation data cannot describe the whole organization.
AI check, no web:Claude Opus:matched our checkGemini Flash:matched our check
17:57
The cyber arms race and data centers
No checkable claims from this segment.
45:22
Economic data
45:571
Finding 1 of 3 ·
45:57–48:58 ·
David Sacks
Seven claims in three minutes: every figure matches its release, and three of the frames don't hold
Between 45:57 and 48:58, David Sacks made seven claims to argue that the economy is "running hot". Each figure matches the agency release it came from. Three of the frames around them do not hold. Unemployment is not "near record lows". The "average" income growth for this term is one year of data. And not every poverty measure is at a historic low. Three more need context. Core inflation has been flat since June once revisions are counted. The August jobs figure was cut on the day the episode aired. And the Journal's comparison counts AI spending projected through 2032.
Q2 GDP was revised up to 2.2% from 1.5%, and Q1 to 2.5% from 2.1%
“Q2 GDP was revised up to 2.2%, up from 1.5%. So that's basically 50% increase. Economists had expected no revision. Q1 was also revised up to 2.5% from 2.1%.”
What the record shows
BEA's third estimate put Q2 2026 real GDP growth at 2.2% annualized, up 0.7 points from the 1.5% second estimate. The market forecast was 1.5%, meaning no revision. The annual update raised Q1 2026 from 2.1% to 2.5%. Later in the episode (52:40), Sacks says "2.5% in Q2", mixing up the two quarters.
August payrolls "just came out" at +162,000 against 55,000 expected
“August payrolls just came out plus 162,000. The consensus was only 55,000, which is a beat of about 107,000 jobs. June and July were also revised up by combined 55,000.”
What the record shows
BLS reported +162,000 for August against a 55,000 consensus, with June and July revised up a combined 55,000. The report came out on Sept 4, four weeks before the episode. On Oct 2, the day the episode was published, BLS revised August down to +133,000 and cut July and August combined by 60,000.
“So unemployment is still near record lows at 4.1% even though labor participation rose .2% to 61.6%.”
What the record shows
August unemployment was 4.1% and participation rose from 61.4% to 61.6%, so the numbers are right. "Near record lows" is not: the record is 2.5% (1953), the low since 1970 is 3.4% (April 2023), and the rate was 4.0% when the term began in January 2025. BLS also notes participation is down 0.5 point since January 2026, and unemployment rose to 4.2% in September.
Core PCE came in at 3.0% against 3.3% expected, so inflation "is cooling"
“We saw that core PCE just came in at 3.0% versus 3.3% expected. So inflation is still a bit too high, but actually it is cooling and beating expectations.”
What the record shows
August core PCE was 3.0% over 12 months, against a 3.3% forecast. The same release carried BEA's annual revision, which cut July from the 3.3% first published to 3.0%. On the revised series, core inflation was 3.2% in May and 3.0% in June, July and August: down 0.2 points since May, and flat since June. Most of the "beat" came from revised history. Headline PCE was 3.4%.
After-tax household income has averaged 3.1% a year in Trump's second term, against 3.9% in his first and −1.2% under Biden
“After tax household income is growing again in the second Trump term. It's averaged 3.1% a year in Trump 47. It was 3.9% in Trump 45 and it was minus 1.2% in Biden's term as a 4-year average.”
What the record shows
All three numbers can be rebuilt from Census's real median post-tax household income series: 2016→2020 = 3.9%/yr, 2020→2024 = −1.2%/yr, 2024→2025 = 3.1%. But the "average" for the second term is a single year. The 2020 endpoint includes stimulus checks, which Census counts in post-tax income. Post-tax income rose that year even though pretax income fell. That raises Trump's first-term figure and lowers Biden's. Using 2019 instead, 2019→2024 is about flat, not −1.2%.
Poverty is at "its lowest level in history, 10.2%", and all poverty rates are at historic lows
“And the poverty rate in the US has fallen to its lowest level in history, 10.2%. So, you know, the poverty rate, the child poverty rate, these are all at historic lows.”
What the record shows
The official poverty rate fell to 10.2% in 2025, the lowest in a series that starts in 1959. Census says it is not statistically different from the 10.5% of 2019. Official child poverty hit a record low of 13.4%. On Census's broader Supplemental Poverty Measure, which counts taxes and benefits, poverty was 13.1%, little changed from 2024 and well above its 2021 low. So "all at historic lows" does not hold.
The Wall Street Journal said the AI build-out is bigger than the railroads, canals and grid combined
“The infrastructure buildout that again the Wall Street Journal said is bigger than the railroads, the canals, and the electric grid combined.”
What the record shows
The WSJ ("The AI Build-Out Is Becoming the Biggest Economic Bet in U.S. History", Sept 24, 2026) compared spending as a share of GDP. Projected AI infrastructure spending of $10.3T from 2025 to 2032 averages about 3.63% of GDP a year. Canals (1836 to 1841, 0.66%), railroads (1870 to 1890, 2.24%) and electrification (1905 to 1925, 0.5%) sum to 3.40%. So "combined" holds. The AI figure, though, is spending projected through 2032, and most of it has not happened yet.
Short-term Treasury yields climbed 60 basis points to their highest since 2002
“We've seen short-term rates on treasuries climb by 60 basis points. That's a pretty consequential climb. I think we haven't seen a short-term yield this high since going back to 2002.”
What the record shows
The 60-basis-point climb holds for short-term rates: the 2-year rose from 4.34% (Aug 28) to 4.92% (Sept 28), and the 10-year from 4.73% to 5.29% (Sept 30). The "highest since 2002" part fits only the 10-year, whose 5.29% close was its highest since May 2002. The 2-year was higher in October 2023 (5.19%) and April 2024 (5.04%).
There were "over 300 days" of Brent crude above $100 under Obama
“By the way, during the Obama presidency, we had over 300 days of Brent crude over $100 a barrel.”
What the record shows
EIA's daily Brent spot series shows 859 trading days above $100 between Jan 20, 2009 and Jan 20, 2017, nearly three times the "over 300" Sacks cited. They fall in 2011 to 2014: 224, 223, 243 and 169 days. For comparison, Trump's second term had 71 such days through Sept 29, 2026.
AI check, no web:Claude Opus:matched our checkGemini Flash:matched our check
58:242
Finding 2 of 3 ·
58:24 ·
Jason Calacanis
Inflation did go into the twos, for nine months of 2025
Jason Calacanis said inflation under Trump has been "never into the twos". BLS figures put headline CPI between 2.3% and 2.9% in nine months of 2025, and at 2.4% in January and February 2026. It rose above 3% in March 2026. Core CPI has stayed between 2.4% and 2.9% every month of 2026.
Headline CPI, 12-month change
Source: BLS, CPI-U all items (series CUUR0000SA0). BLS published no October 2025 figure, so the line has a gap there.
Inflation under Trump has been "never into the twos"
“If you look at unemployment, hasn't changed. It's about the same. And then if you look at inflation, it's still sustained at that high 3s, sometimes dipping into four, sometimes as low as 3.2, but never into the twos.”
What the record shows
Headline CPI inflation was in the 2s for most of Trump's second term: between 2.3% and 2.9% in nine months of 2025, and 2.4% in January and February 2026. It rose above 3% from March 2026, peaked at 4.2% in May, and ran at 3.4% in July and August. Core CPI has been between 2.4% and 2.9% every month of 2026. The unemployment half of his point holds: 4.0% in January 2025 and 4.1% in August 2026.
AI check, no web:Claude Opus:matched our checkGemini Flash:couldn't check
1:00:24
The midterms
1:08:573
Finding 3 of 3 ·
1:08:57 ·
David Friedberg
The middle class holds about 29% of U.S. wealth, not 87%
David Friedberg said $160 trillion of America's $183 trillion in net worth sits with the middle class. That would be about 87%. The Federal Reserve's distributional accounts put the 50th to 90th percentile at $53.4 trillion in Q2 2026, about 29%. The top 10% holds about 69%. Even counting everyone below the top 0.1% as middle class reaches only about $156 trillion to $158 trillion. His case that only middle-class taxes, deep spending cuts or entitlement reform can close the deficit rests on the larger share. The Fed does not define "middle class". The 50th to 90th percentile is the band it reports that comes closest.
As claimed
Middle class $160T
Fed, Q2 2026
50th to 90th $53.4T
Top 10% $128T
Source: Federal Reserve, Distributional Financial Accounts, Q2 2026. The bottom 50% held $4.3 trillion, the 50th to 90th percentile $53.4 trillion and the top 10% about $128 trillion, against the claimed $160 trillion of $183 trillion.
$160 trillion of America's $183 trillion net worth "sits" with the middle class
“You either need to tax the middle class, which is where 160 trillion of America's $183 trillion net worth sits, or you've got to cut spending massively.”
What the record shows
$183T matches the Fed's Q1 2026 household net worth ($195.9T in Q2 after a $12.8T rise). Most of it sits with the top 10%. In Q2 2026 the 50th to 90th percentile held $53.4T (≈29%), the bottom 50% $4.3T (≈2%), and the top 10% about $128T (≈69%). Even everyone below the top 1% held about $125T, not $160T. Only by counting everyone below the top 0.1% as middle class does the total reach about $156T to $158T.
The New York Times, WSJ, BBC, Sky, AFP and CNN described the flydubai hijacking as an altercation, a struggle, a fight or a brawl
“The first one was New York Times came out and said it was an altercation and then the Journal called it a struggle. The BBC called it a fight between pilots. Sky News called it a brawl between pilots. AFP called it, you know, said they were fighting. Even CNN said it was a altercation.”
What the record shows
Confirmed for four outlets. NYT posted that the plane 'was diverted after an altercation between the pilot and co-pilot, Israeli officials said'. BBC ran 'Israel-bound flight diverted after "fight" between pilots'. Sky ran 'brawl between pilots'. AFP said 'a fight had broken out between the plane's pilots'. Not found: WSJ's earliest headline cited a 'steep dive', and CNN's first story said 'violent incident'. Most early wording repeated Israeli officials and flydubai's own statement that 'an altercation occurred in the flight deck'.
AI check, no web:Claude Opus:couldn't checkGemini Flash:couldn't check
Chamath Palihapitiya made no claim we could check in this episode.
How we check
We take the claims that carry the argument and check each one against primary sources: the agency release, the filing, the original article. We link every source with its date. The speaker carries the burden of proof, so a claim with no evidence either way is rated Unsupported, not false. We check in both directions, and we say so when a speaker understated their own case.
The rating scale
Accurate
The claim is right as stated.
Mostly accurate
Right in substance, though a detail is off or out of date.
Missing context
The figure is right, but the frame around it leaves a false impression.
Misleading
Built from real data, but arranged to mean something the data does not show.
Inaccurate
Wrong: the record shows a different figure or fact.
Unsupported
We found no evidence for it, and none against it.
The headline counts group the six ratings into three. Hold up covers Accurate and Mostly accurate. Need context covers Missing context and Misleading. Don't hold up covers Inaccurate and Unsupported.
How this page was made
Every claim is checked against primary sources, and every verdict gets an independent second review before we publish. We use AI to research and draft, and no verdict rests on a model's memory: each one cites a source you can open. When a claim is about an AI company, a model made by that company does not get the final say.
What the AI check shows
On October 7, 2026, we gave every claim to Claude Opus and Gemini Flash cold, with no web access, the way a reader might paste it into a chatbot. We scored each answer against our check and show the result at the foot of each claim card. It shows what a chatbot alone would have told you and plays no part in our verdicts.
Claude Opus: 7 matched, 5 couldn't check, 1 confidently wrong.
BLS revised August payrolls from +162,000 to +133,000 on October 2, 2026, the day the episode was published. Our verdict takes both figures into account.
Corrections
None yet.
Found an error? Email [email protected] with the claim and your source. We log every correction on the page, with its date.