A study by Convovo, an AI research studio led by Cornell professor Tapan Parikh, asked two AI models to name one candidate in every 2026 House, Senate and governor race, with no hedging. It used GPT-6.1 Sol and Grok 4.7 through their APIs, with web search on. GPT chose the Democrat in 1,082 of 1,111 answers naming a major-party candidate. Grok chose the Republican in 952 of 1,038. TechCrunch said the two agreed under 10% of the time across nearly 500 races.
Parikh told TechCrunch the findings show how the models decide what to search for. Convovo logged 9,484 searches for Grok and 1,872 for GPT. TechCrunch said Grok's results often included a Christian conservative voter guide, Turning Point Action and X. GPT's leaned on Wikipedia, the AP, Axios and Reddit. Convovo also says xAI adds about 1,240 tokens of hidden instructions to each Grok call, which it cannot see. The apps decline by default, but Parikh told TechCrunch simple prompts got them to recommend candidates, mostly in line with his earlier research. OpenAI said the API test does not reflect ChatGPT. Neither source says why the models lean.
Teams building on these APIs should ask who controls the system prompt, which sites the search returns, and what happens when the API answers a question the app would decline.