Politico reports that it was sent communications from inside the Trump campaign, including Sen. J.D. Vance’s (R-OH) 271-page vetting file, allegedly by an Iranian hacker.

The outlet said that it has been receiving anonymous emails containing internal communications from the Trump campaign. The campaign acknowledged the authenticity of the communications on Saturday, accusing “foreign sources hostile to the United States,” for leaking them.

“These documents were obtained illegally from foreign sources hostile to the United States, intended to interfere with the 2024 election and sow chaos throughout our Democratic process,” Trump spokesperson Steven Cheung told Politico. “On Friday, a new report from Microsoft found that Iranian hackers broke into the account of a ‘high ranking official’ on the U.S. presidential campaign in June 2024, which coincides with the close timing of President Trump’s selection of a vice presidential nominee.”

“Another Iranian group, this one connected with the Islamic Revolutionary Guard Corps, or IRGC, sent a spear phishing email in June to a high-ranking official on a presidential campaign from the compromised email account of a former senior advisor,” the Microsoft report stated. “The email contained a link that would direct traffic through a domain controlled by the group before routing to the website of the provided link.”

“Within days of this activity, the same group unsuccessfully attempted to log into an account belonging to a former presidential candidate,” the report added.

Politico reported that on July 22 it began receiving emails from an AOL account only identified by the moniker “Robert” that contained internal communications from the Trump campaign.

One email contained a 271-page report on Vance entitled “POTENTIAL VULNERABILITIES.” Robert told Politico they had a “variety of documents from [Trump’s] legal and court documents to internal campaign discussions,” and when pressed for how they obtained them, Robert responded, “I suggest you don’t be curious about where I got them from. Any answer to this question, will compromise me and also legally restricts you from publishing them.”

This is by no means a first for Trump. In 2020, a hacker hacked his X account by guessing his password, ‘maga2020!’

Trump has also allegedly been the target of an Iranian assassination plot which resulted in a Pakistani man named Asif Merchant, with ties to the Islamic Republic, being charged by the DOJ on Wednesday.

How This Poll Was Conducted

Here are the key things to know about these Times/Siena polls:

• Interviewers spoke with 619 registered voters in Michigan and 661 registered voters in Wisconsin from Aug. 5 to 8, and 693 registered voters in Pennsylvania from Aug. 6 to 9.

• Times/Siena polls are conducted by telephone, using live interviewers, in both English and Spanish. About 90 percent of respondents were contacted on a cellphone for these polls.

• Voters are selected for Times/Siena surveys from a list of registered voters. The list contains information on the demographic characteristics of every registered voter, allowing us to make sure we reach the right number of voters of each party, race and region. For this poll, interviewers placed more than 237,000 calls to nearly 118,000 voters.

• To further ensure that the results reflect the entire voting population, not just those willing to take a poll, we give more weight to respondents from demographic groups that are underrepresented among survey respondents, like people without a college degree. You can see more information about the characteristics of our respondents and the weighted sample at the bottom of the page, under “Composition of the Sample.”

• The margin of sampling error among registered voters is plus or minus 4.5 percentage points in Michigan, plus or minus 4 percentage points in Pennsylvania, and plus or minus 4.3 percentage points in Wisconsin. In theory, this means that the results should reflect the views of the overall population most of the time, though many other challenges create additional sources of error. When computing the difference between two values — such as a candidate’s lead in a race — the margin of error is twice as large.

If you want to read more about how and why we conduct our polls, you can see answers to frequently asked questions and submit your own questions here.

Full Methodology

The New York Times/Siena College polls in Michigan and Wisconsin were conducted in English and Spanish on cellular and landline telephones from Aug. 5 to 8, 2024, and from Aug. 6 to 9, 2024, in Pennsylvania. In all, 1,973 registered voters were interviewed. When all states are joined together, the margin of sampling error is plus or minus 2.5 percentage points for all registered voters and plus or minus 2.6 percentage points for the likely electorate.

The margin of sampling error among registered voters for each state poll is plus or minus 4.5 percentage points in Michigan, plus or minus 4 points in Pennsylvania and plus or minus 4.3 points in Wisconsin. Among the likely electorate, it is plus or minus 4.8 percentage points in Michigan, plus or minus 4.2 points in Pennsylvania and plus or minus 4.3 points in Wisconsin.

Sample

The survey is a response rate-adjusted stratified sample of registered voters on the L2 voter file. The sample was selected by The New York Times in multiple steps to account for differential telephone coverage, nonresponse and significant variation in the productivity of telephone numbers by state.

The L2 voter file for each state was stratified by statehouse district, party, race, gender, marital status, household size, turnout history, age and homeownership. The proportion of registrants with a telephone number and the mean expected response rate, based on prior Times/Siena polls, were calculated for each stratum. The initial selection weight was equal to the reciprocal of a stratum’s mean telephone coverage and modeled response rate. For respondents with multiple telephone numbers on the L2 file, the number with the highest modeled response rate was selected.

Fielding

The samples for each state were stratified by political party, race and region and were fielded by the Siena College Research Institute, with additional field work by ReconMR, the Public Opinion Research Laboratory at the University of North Florida and the Institute of Policy and Opinion Research at Roanoke College. Interviewers asked for the person named on the voter file and ended the interview if the intended respondent was not available. Overall, 90 percent of respondents were reached on a cellular telephone.

The instrument was translated into Spanish by ReconMR. Bilingual interviewers began the interview in English and were instructed to follow the lead of the respondent in determining whether to conduct the survey in English or Spanish. Monolingual Spanish-speaking respondents who were initially contacted by English-speaking interviewers were recontacted by Spanish-speaking interviewers. Overall, 13 percent of interviews among self-reported Hispanics were conducted in Spanish, including 13 percent of weighted interviews.

An interview was determined to be complete for the purposes of inclusion in the ballot test question if the respondent did not drop out of the survey by the end of the two self-reported variables used in weighting — age and education — and answered at least one of the age, education, race or presidential election ballot test questions.

Weighting — registered voters

The survey was weighted by The Times using the R survey package in multiple steps.

First, the sample was adjusted for unequal probability of selection by stratum.

Second, the sample was weighted to match voter file-based parameters for the characteristics of registered voters.

The following targets were used:

• Party (party registration if available in the state, else classification based on participation in partisan primaries if available in the state, else classification based on a model of vote choice in prior Times/Siena polls) by whether the respondent’s race is modeled as white or nonwhite (L2 model) in Michigan, whether the respondent voted in 2020 in Wisconsin, and the voter file age of the respondent in Pennsylvania.

• Age (Self-reported age, or voter file age if the respondent refuses) by gender (L2)

• Race or ethnicity (L2 model)

• Education (four categories of self-reported education level, weighted to match NYT-based targets derived from Times/Siena polls, census data and the L2 voter file)

• White/non-white race by college or non-college educational attainment (L2 model of race weighted to match NYT-based targets for self-reported education)

• Marital status (L2 model)

• Home ownership (L2 model)

• State region (NYT classifications)

• Turnout history (NYT classifications based on L2 data)

• Method of voting in the 2020 elections (NYT classifications based on L2 data)

• The average recalled 2020 vote choice of respondents to the last three New York Times/Siena College polls, including the results of this poll before the recalled vote weight, in Wisconsin.

Finally, the sample of respondents who completed all questions in the survey was weighted identically, as well as to the result for the general election horse race question (including leaners) on the full sample.

Weighting — likely electorate

The survey was weighted by The Times using the R survey package in multiple steps.

First, the samples were adjusted for unequal probability of selection by stratum.

Second, the first-stage weight was adjusted to account for the probability that a registrant would vote in the 2024 election, based on a model of turnout in the 2020 election.

Third, the sample was weighted to match targets for the composition of the likely electorate. The targets for the composition of the likely electorate were derived by aggregating the individual-level turnout estimates described in the previous step for registrants on the L2 voter file. The categories used in weighting were the same as those previously mentioned for registered voters.

Fourth, the initial likely electorate weight was adjusted to incorporate self-reported intention to vote. Four-fifths of the final probability that a registrant would vote in the 2024 election was based on their ex ante modeled turnout score and one-fifth based on their self-reported intentions, based on prior Times/Siena polls, including a penalty to account for the tendency of survey respondents to turn out at higher rates than nonrespondents. The final likely electorate weight was equal to the modeled electorate rake weight, multiplied by the final turnout probability and divided by the ex ante modeled turnout probability.

Finally, the sample of respondents who completed all questions in the survey was weighted identically, as well as to the result for the general election horse race question (including leaners) on the full sample.

The margin of error accounts for the survey’s design effect, a measure of the loss of statistical power due to survey design and weighting. The design effect for the full sample is 1.17 for registered voters and 1.26 for the likely electorate in Pennsylvania, 1.25 for registered voters and 1.28 for the likely electorate in Wisconsin, and 1.33 for registered voters and 1.46 for likely voters in Michigan.

For the sample of completed interviews, the margin of sampling error is plus or minus 5.3 points for registered voters and plus or minus 5.5 points for the likely electorate in Michigan, plus or minus 4.5 points for registered voters and plus or minus 4.6 points for the likely electorate in Pennsylvania, and plus or minus 4.7 points for both registered voters and the likely electorate in Wisconsin.The design effect for the sample of completed interviews is 1.24 for registered voters and 1.33 for the likely electorate in Pennsylvania, 1.3 for registered voters and 1.32 for the likely electorate in Wisconsin, and 1.46 for registered voters and 1.57 for the likely electorate in Michigan.

Historically, The Times/Siena Poll’s error at the 95th percentile has been plus or minus 5.1 percentage points in surveys taken over the final three weeks before an election. Real-world error includes sources of error beyond sampling error, such as nonresponse bias, coverage error, late shifts among undecided voters and error in estimating the composition of the electorate.

Just two of OpenAI’s 11-strong founding team are still active at the ChatGPT maker, after an exodus following November’s attempted boardroom coup against chief executive Sam Altman.

Three co-founders have departed so far this year, including John Schulman, who defected to its artificial intelligence rival Anthropic this week. Greg Brockman, OpenAI’s president, also said on Monday he would be taking extended leave from the company.

A high rate of turnover is not unusual at a start-up. However, attrition of senior figures at OpenAI has stepped up in recent months following November’s leadership crisis, when Altman was fired by his board only to be reinstated days later.

Since then, the loss of executives and staffers working on AI safety and research has raised questions about the direction of the $86bn company, which is in a fierce battle to stay ahead of rivals including Google and Anthropic.

Some co-founders have absconded to rivals, others to launch their own AI companies, while the team’s most famous former member — Elon Musk — has become a vociferous critic of OpenAI in public and in the courts.

OpenAI had a larger number of founders than most Silicon Valley start-ups because Altman and Brockman wanted to build an AI supergroup of the top researchers in the field when it got started in 2015. Here is where those 11 founders are now.

Leavers

Greg Brockman

on a leave of absence since August 2024

Brockman is a core member of OpenAI’s founding team. He was persuaded by Altman and Musk to leave his job as chief technology officer at financial technology company Stripe and take on the same position at OpenAI.

He has been a key Altman ally since the beginning. When the board moved against Altman in a coup in November, Brockman was also removed as a director. The two returned to their posts together when the board backtracked five days later.

On Monday, the company’s president announced he would be taking a sabbatical for the rest of the year.

“First time to relax since co-founding OpenAI 9 years ago,” he wrote on X. “I’ve poured my life for the past 9 years into OpenAI, including the entirety of my marriage. Our work is important to me, but so is life.”

John Schulman

joined Anthropic in August 2024

Schulman, a research scientist who played a vital role in building the company’s ChatGPT chatbot, announced he would leave OpenAI on Monday. He was responsible for fine-tuning the company’s AI models and ensuring they behaved in a way that conformed to human values — a process known as alignment.

He will take up a similar role at rival start-up Anthropic, which itself was founded by ex-OpenAI researchers in 2021.

“This choice stems from my desire to deepen my focus on AI alignment, and to start a new chapter of my career where I can return to hands-on technical work, alongside people deeply engaged with the topics I’m most interested in,” Schulman said in a note to colleagues on Monday.

Ilya Sutskever

left to found Safe Superintelligence in May 2024

Sutskever left his position as OpenAI’s chief scientist six months after voting with the company’s board to remove Altman. Sutskever, one of the most prominent researchers in the field, reversed his position and backed the chief executive’s return a few days later.

Nonetheless, he has been largely absent from public view in the months since the abortive coup, and in May he left to start a company called Safe Superintelligence.

Andrej Karpathy

left to found Eureka Labs in February 2024

Karpathy, a research scientist who was advised at Stanford by “Godmother of AI” Fei-Fei Li, first left OpenAI in 2017 to join Tesla as a senior director. He returned to OpenAI in 2023 and left again a year later to launch Eureka Labs, which is building AI teaching assistants.‎

Durk Kingma

left for Google Brain in June 2018

Kingma, who worked on developing algorithms for generative AI models, left for Google in the summer of 2018. He has continued to lead research on large language models and image models at Google Brain, which merged with DeepMind last year.

Elon Musk

resigned from the board in 2018

Musk, who provided much of OpenAI’s early funding, left the company in 2018 after clashing with Altman over the direction of research. The billionaire launched a rival company, xAI, last year and claims he can overhaul OpenAI’s lead.

The Tesla, SpaceX and X chief has also launched a number of lawsuits against Altman and OpenAI, arguing this week that he was induced to invest in the AI company by its “fake humanitarian mission”.

Pamela Vagata

JOINED STRIPE in 2016

Vagata, listed as a founding member of OpenAI in the company’s launch announcement, makes no mention of the start-up in her LinkedIn profile. She joined Stripe as a technical leader in the fintech company’s AI team in 2016, and founded early-stage venture capital firm Pebblebed in 2021.

‎Vicki Cheung

JOINED LYFT in 2017

Cheung, who worked on language-learning app Duolingo before becoming OpenAI’s first engineer, left the company in 2017 to join ride-hailing start-up Lyft. In 2020 she founded machine learning start-up Gantry alongside former OpenAI researcher Josh Tobin.

Trevor Blackwell

left in 2017

Blackwell was a partner at Y Combinator, the San Francisco start-up accelerator that Altman ran before establishing OpenAI. He helped launch the AI company and left in 2017. A robotics enthusiast, he is now based in Gloucestershire, England.

‎Remainers

Sam Altman

Altman remains as OpenAI’s chief executive after surviving a boardroom coup in November, during which directors accused him of not being “consistently candid” with them. He was reinstated five days after being fired on the back of a campaign by employees and investors in OpenAI, including Microsoft.

The departure of other senior figures has left the 39-year-old as by far the most prominent figure at the company, and the reconstitution of the board after its failed ousting has further solidified his power.

Wojciech Zaremba

Polish computer scientist Zaremba remains at OpenAI where he works as a researcher. He called on the board to resign after they moved against Altman, and has since urged his chief executive and Musk to drop their “unnecessary fight”.

“It would be so much better to put your creative energy into building the future you dream of over a quarrel. May you (both) be happy and find peace,” he wrote in a post on X in March, signing off with a love heart.

Additional reporting by Madhumita Murgia