OpenAI Expenses in Campaigns: 39 US Candidates Paid for ChatGPT
It only takes twenty dollars a month for a paid version of ChatGPT, yet that seemingly minimal expense has turned into a political issue. An analysis by the Washington Post of documents filed with the Federal Election Commission has revealed OpenAI's expenses in US electoral campaigns this cycle: thirty-nine congressional candidates reported payments to OpenAI, the company behind ChatGPT, for the use of artificial intelligence tools in their campaigns. A number that, according to those closely following the sector, tells only part of the story.
Summary
- Key Points
- Extensive use of OpenAI tools in congressional campaigns
- Who spends and how much
- Violations of rules and use beyond OpenAI's guidelines
- OpenAI's ban on political advertising
- Synthetic media and personalized communications
- Transparency and regulatory gaps in AI campaign spending
- Outdated spending categories
- Why the real numbers are likely higher
- FAQ
- How many congressional candidates reported expenses for OpenAI subscriptions?
- Does OpenAI allow its tools to be used for creating political ads?
- Why is it difficult to accurately monitor campaign spending on AI tools?
- Do all campaigns fully disclose their use of AI tools in their reports?
Key Points {#Key_Points}
- Thirty-nine congressional candidates reported payments to OpenAI in this election cycle, according to the FEC document analysis.
- Republican Representative Mike Lawler spent about $3,260 on OpenAI services since 2025.
- The Republican National Committee has paid nearly $9,700 in OpenAI subscriptions, the highest amount among committees.
- About 30 PACs and political parties reported similar expenses.
- At least two candidates used the tools to produce political ads, violating OpenAI's usage rules.
Extensive Use of OpenAI Tools in Congressional Campaigns {#Extensive_Use_of_OpenAI_Tools_in_Congressional_Campaigns}
Generative artificial intelligence has become a recurring item in campaign budgets. Documents filed with the FEC show that the use of ChatGPT and other OpenAI tools is no longer an isolated exception, but a widespread practice among candidates, committees, and political organizations at all levels.
Who Spends and How Much {#Who_Spends_and_How_Much}
Among individual candidates, New York Republican Mike Lawler leads the individual spending chart with about $3,260 paid to OpenAI since 2025. On the organizational front, the Republican National Committee is the largest contributor with payments nearing $9,700, the highest amount recorded among political action committees. Additionally, about 30 PACs and political parties reported similar payments to OpenAI during the election cycle.
These numbers, it should be noted, represent only what the campaigns have chosen to voluntarily disclose in their reports. Campaign staff may pay for personal subscriptions, use free versions of the services, or route payments through vendors and consultants whose invoices do not specify which tools are actually being used.
Violations of Rules and Use Beyond OpenAI's Guidelines {#Violations_of_Rules_and_Use_Beyond_OpenAI's_Guidelines}
Not all declared expenses fall within the permitted use. At least two campaigns have used their OpenAI subscriptions specifically to generate political advertising content, a use that is in open contradiction to the company's policies.
OpenAI's Ban on Political Advertising
OpenAI explicitly prohibits the use of its tools to produce content intended for political advertising, a restriction introduced well before the start of this electoral cycle. Despite this, at least two candidates have decided to consider the terms of service more as a suggestion than as a binding rule. It should be noted that OpenAI does not have officials dedicated to monitoring political compliance who oversee every conversation initiated on ChatGPT: the verification of policy compliance is therefore almost exclusively reliant on the self-reporting of the involved parties.
Synthetic Media and Personalized Communications
A third candidate has pushed the use beyond mere back-office efficiency, claiming to have employed unspecified artificial intelligence tools to draft communications, personalize messages, and produce what the Washington Post describes as synthetic media. This raises non-trivial questions about what actually reaches voters without any label indicating its artificial origin.
Transparency and Regulatory Gaps in AI Campaign Spending
Federal regulations on electoral funding have yet to catch up with generative artificial intelligence. The law requires the declaration of expenses, but the available categories for reporting were designed for a world of printing costs, travel, and staff salaries: an "OpenAI subscription" does not have its own dedicated budget line.
Outdated Spending Categories
This gap in reporting categories allows campaigns to legitimately file these expenses under generic headings, which end up obscuring the true nature of the expenditure. The Federal Election Commission has long been discussing whether to impose a requirement to declare content generated with artificial intelligence, but as of mid-2026, no binding federal rule has yet been finalized. Some states have moved more quickly, passing laws that require explicit notices on political advertising generated with AI or banning the creation of synthetic media depicting candidates.
Why the Real Numbers Are Likely Higher
Political consultants emphasize that many campaigns prefer to keep a low profile on the use of AI to avoid voter skepticism towards artificially generated content. The count of thirty-nine candidates made by the Washington Post should therefore be interpreted more as a minimum threshold than as a definitive estimate. The leap from the early years of public availability of ChatGPT, between late 2022 and 2023, when the use of AI in campaigns was mostly anecdotal, is nonetheless significant. And without binding disclosure obligations, campaigns have no duty to report which content has been drafted, refined, or entirely produced by a language model.
FAQ
How many congressional candidates have declared expenses for OpenAI subscriptions?
Thirty-nine candidates for Congress have reported payments to OpenAI for the use of artificial intelligence tools during this election cycle.
Does OpenAI allow the use of its tools to create political advertisements? {#OpenAI_allows_use_of_its_tools_to_create_political_advertisements}
No, OpenAI prohibits the use of its tools to produce political advertising content, but at least two candidates have violated this rule.
Why is it difficult to accurately monitor campaign spending on AI tools? {#Why_is_it_difficult_to_monitor_campaign_spending_on_AI_tools}
Campaign finance disclosures do not include a dedicated category for artificial intelligence tools, so expenses are often recorded under generic headings that obscure their true nature.
Do all campaigns fully disclose their use of AI tools in their reports? {#Do_all_campaigns_fully_disclose_their_use_of_AI_tools_in_their_reports}
No, political consultants report that many campaigns use personal accounts or vague spending categories to completely hide payments related to artificial intelligence tools.
Content created with the assistance of artificial intelligence and human editorial review.
-- Price
This content is provided for general informational purposes only and doesn't constitute financial, investment, legal, or tax advice. Any events, rewards, online promotions, or related information mentioned herein should not be considered a recommendation, solicitation, or invitation to purchase, sell, trade, or otherwise deal in any crypto assets. Crypto assets are highly volatile and may result in loss. The availability of WEEX services, products, and related events may vary by region. You are responsible for ensuring that your participation is in accordance with applicable local laws and regulations.
You may also like

Three Major Lending Protocols Enter Fixed Rate Market: What Innovations Do They Bring?

Banca d’Italia demands checks on every crypto transfer

MoonPay CEO Asks Claude to Book a Flight Before Stuffing a Crypto Wallet into the Chatbox

Poland's crypto licensing deadlock benefits EU firms

HashKey Cloud joins Stacks Bitcoin staking launch

Reviews of awx pro: what the crypto exchange offers, how trading works, and where the risks arise

Does the massive return of leverage threaten the recent rally?

Australia tightens crypto oversight with 45 removals

Strong jobs just triggered a hawkish UBS reversal that could pressure Bitcoin through December

Solana beats Bitcoin on one key metric, but a single software bug could still take down the network

Research Indicates Critical Threshold for Stablecoin Decoupling

Ethereum's Vitalik Buterin puts 60% odds on a cryptography breakthrough that could weaken Wall Street middlemen

Polymarket Report for the First Half of 2026: High-Frequency Traders or Super Forecasters?

Meta Creates AI Filter to Select Experiments Before Spending GPU Hours

TRON Industry Weekly Report: Non-Farm Payrolls Lean Hawkish but CPI Will Determine If BTC Can Hold Above 80,000, Detailed Analysis of KOR Protocol for Digital Content and IP Assetization

Wealthy Tax Avoidance: DeFi Lending Pools Take the Blame

唐华斑竹: USDD's Eighth Strategy Event Launched with a Total Prize Pool of $700,000

Raydium LaunchLab adds support for any token pair on Solana

Bercy Plans to Subject Employee Savings to Social Contributions in 2027

Altcoin open interest overtakes Bitcoin after 21 months

Atomic Time Faces Historic Change to Avoid Negative Leap Second

How Bitmine could surpass its 5% Ethereum goal without buying more ETH

OpenAI's GPT-6 Astra Is Shockingly Good at Almost Everything

Poland: Parliament Fails to Overturn Veto on Crypto Law

Solana App Revenue Hits $143 Million: The Best Month Ever?

USDC Stablecoin Growth: Dominates On-Chain Volumes, +$1 Billion Market

Cryptocurrency Regulation in Russia Has Started, but the Market Needs More Time

TRON Account Growth Reaches 402 Million: Can It Challenge Ethereum in the Stablecoin Market?

Threshold to Break a Bitcoin Key Drops to 813 Logical Qubits










