Kalshi develops in-house AI tool to review prediction market contracts

Dig Deeper

The Backstory

AI arrives as contract scrutiny intensifies

Kalshi’s use of an in-house artificial intelligence tool to help review prediction market contracts reflects a broader shift in the company’s business: The exchange is scaling quickly while facing rising pressure to prove its markets are clear, enforceable and resistant to abuse. Contract wording is central to that challenge. In event markets, small differences in phrasing can determine whether a contract settles cleanly or creates disputes over what outcome actually occurred.

The tool, known internally as Harrison, is being used as Kalshi expands across political, sports and entertainment-related markets. Its role in reviewing templates and identifying potential weaknesses comes as the company says it has created about 2,800 market templates. That industrial approach to market creation is important because prediction exchanges rely on standardized contracts that can be adapted quickly to news cycles and major events.

The timing also matters. Kalshi recently said trading volume increased 800% over several months and that it accounts for about 90% of U.S. prediction market activity. With that growth has come a larger regulatory target. Lawmakers and regulators are now focused not only on whether prediction markets should exist at scale, but also on how platforms police insider information, manage conflicts and distinguish financial event contracts from gambling products.

From election contracts to sports expansion

Kalshi’s current position was shaped by its earlier fight with the Commodity Futures Trading Commission over political event contracts. The company challenged the agency after seeking to offer markets tied to the U.S. presidential election. Kalshi ultimately won approval for election event contracts, and those markets drew national attention during the 2024 election cycle as users and analysts compared their signals with polling and other forecasting tools.

That legal and commercial breakthrough helped establish Kalshi as the leading U.S. regulated prediction market, but it also raised the stakes for every subsequent product decision. Political contracts demonstrated both the appeal and the sensitivity of event markets. They also created a template for the argument Kalshi and similar companies continue to make: that federally regulated event contracts are financial instruments overseen by the CFTC, not state-regulated wagers.

The company then pushed deeper into sports. In January, Kalshi launched event contracts on major sports outcomes, including formats asking whether a team would win a title. The move put the company in direct competition with sportsbooks and other event-contract providers. It also moved prediction markets into an area with a long-established state gambling framework, making conflict with state regulators more likely.

Sports contracts widened Kalshi’s addressable market, but they also complicated its regulatory posture. Political markets had already drawn scrutiny because public officials and government employees may possess nonpublic information. Sports markets raised different concerns, including manipulation risks and overlap with prop-style betting. The company’s use of AI to stress-test wording and templates should be understood against that backdrop: Market scale now depends on legal defensibility and operational discipline, not just user demand.

Regulators seek a national framework

The CFTC has been trying to define its role as prediction markets grow. A proposal on regulating these markets is now under White House review, according to a filing reported earlier by Bloomberg and covered in a report on the CFTC prediction market proposal. Details have not been disclosed, but the review signals that the federal government is moving toward a more formal framework for markets tied to sports, politics and other public events.

The agency’s position has generally been that it has exclusive authority over federally registered prediction market exchanges. That position is contested. Several states have argued that sports-related contracts offered to retail users function like gambling and should be subject to state law. New York Attorney General Letitia James has sued prediction market platforms, Minnesota enacted a state-level ban and Illinois has moved to restrict certain wagering activity involving state officials.

The divide has practical consequences. If the CFTC’s authority is upheld broadly, prediction exchanges could operate nationwide under one federal regime. That would preserve a major advantage over traditional sports betting, which remains restricted state by state. If states prevail, operators could face a patchwork similar to online sportsbooks, reducing the value of federal registration and forcing exchanges to modify or withdraw products in some jurisdictions.

The CFTC has also issued guidance to prediction markets on sports event contracts, warning exchanges to assess whether certain products are vulnerable to manipulation. The agency specifically highlighted markets based on the actions or performance of a single player, because one person may be able to influence the outcome. That concern mirrors long-running integrity issues in sports betting, but prediction markets add a financial-market structure and federal oversight layer.

Insider trading concerns move to Congress

Federal scrutiny is not limited to agency rulemaking. House Oversight and Government Reform Committee Chairman James Comer has said the committee will investigate potential insider trading on Kalshi and Polymarket. The inquiry, detailed in coverage of the House committee’s probe into prediction market insider trading, seeks information on identity verification, geographic restrictions and systems used to detect unusual trading activity.

The congressional focus reflects a core vulnerability for prediction markets tied to government action. Public officials, congressional staff, administration employees, military personnel and contractors may learn information before it becomes public. If they can trade on that information, prediction markets risk becoming venues for monetizing government access. That is especially sensitive when contracts involve elections, policy decisions, military action or international crises.

The concern is not theoretical. Reports cited by lawmakers have included trades placed before sensitive political and military developments, including activity tied to international conflicts. A U.S. soldier was arrested after allegedly using inside information to trade on political developments in Venezuela through Polymarket. Those cases have amplified calls for stronger surveillance standards and clearer rules on who may trade, when and on which events.

For Kalshi, surveillance and contract design are linked. A poorly framed contract can create ambiguity at settlement. A sensitive contract can invite trading by people with privileged information. A high-volume market can create incentives for manipulation or information leakage. AI tools may help identify drafting issues faster, but lawmakers are likely to judge platforms by outcomes: whether they can detect suspicious trades, prevent prohibited access and provide audit trails when questioned.

Sports economics raise the stakes

The commercial incentive to resolve these issues is substantial. Bank of America has projected that the U.S. market for sports-related event contracts could reach about $1.1 trillion in annual volume. In its analysis of the sports event contracts market, the bank estimated that a 1% average fee could translate into roughly $10 billion in annualized revenue for event-betting companies.

That projection explains why sports has become the central battleground. Kalshi’s sports bets made up 79% of its trading volume in March, according to the Bank of America analysis, while the company held about 90% of the U.S. prediction-market exchange market. Its advantages include federal regulation, availability in all 50 states and access to users 18 and older. Traditional sportsbooks, by contrast, typically operate only in states where sports betting is legal and often serve users 21 and older.

Prediction markets also avoid some costs that weigh on sportsbooks, including state gaming taxes that can consume a large share of revenue. That difference has attracted interest from major gambling operators. DraftKings and FanDuel have introduced prediction-market products, signaling that established sportsbooks see event contracts as both a threat and an opportunity.

The competitive dynamic may accelerate innovation while increasing regulatory friction. If prediction markets can offer sports exposure nationwide with lower tax burdens and younger eligible users, state gambling regulators are likely to challenge them aggressively. If regulators restrict sports event contracts or require state-level compliance, the growth case could narrow. Kalshi’s operational investments, including AI-supported market review, are therefore tied to a larger effort to maintain credibility as the sector moves from niche forecasting to mainstream financialized betting.

A test of scale and legitimacy

The current moment is a test of whether prediction markets can scale without losing regulatory trust. Kalshi’s AI tool addresses one part of that problem by helping review contract language and anticipate flaws before markets go live. But technology alone cannot settle the larger questions now before regulators, courts and Congress.

The industry’s future will depend on whether federal oversight can create enforceable standards for market integrity, consumer protection and insider trading controls. It will also depend on whether prediction exchanges can convince states that sports and political event contracts are not simply gambling by another name. Those disputes will shape the economics of the sector, the products users can access and the compliance obligations operators must meet.

Kalshi’s rapid growth has made it the leading example of both the promise and the risks of prediction markets. Its push into AI-supported contract review shows how the company is trying to manage complexity at scale. The broader question is whether that operational sophistication will be enough as prediction markets move deeper into politics, sports and public policy — areas where information, incentives and regulation collide.