Start free trial

TRUE, BUTLab Not Field

No search: Grok declined 441 races

Turn off search and give it the names, and Grok declined 441 races. The lean sat in the forced prompt and the search.

“Forced to pick, Grok's AI model chooses Republicans and OpenAI's chooses Democrats: Grok 4.7 recommended the Republican in 92% of answers, GPT-6.1 Sol the Democrat in 97%.”

Convovo study led by Cornell professor Tapan Parikh, reported by TechCrunch (24 October 2026)

Recorded 26 October 2026. Our ruling means: literally true, materially misleading.

Read the story →

Receipts pack

The evidence file for this verdict

13 receipts: 4 support the claim, 1 contradict it, 8 add context. 5 figures on the record, 5 traced to a quoted receipt.

The receipts

Each source, the line we quote from it, its date and which way it cuts. Free to read and to cite.

  1. 1Supports the claimconvovo.ai ↗

    We asked GPT-6.1 Sol and Grok 4.7 via the API with search on for every 2026 House, Senate and governor race.

    Look up an archived copy ↗

  2. 2Supports the claimtechcrunch.com ↗

    recommended the Republican candidate in 92% of its answers

    Look up an archived copy ↗

  3. 3Supports the claimraw.githubusercontent.com ↗

    GPT-6.1 Sol names the Democrat in 1,082 of its 1,111 major-party answers (97%)

    Look up an archived copy ↗

  4. 4Supports the claimraw.githubusercontent.com ↗

    Grok 4.7 names the Republican in 952 of 1,038 (92%)

    Look up an archived copy ↗

  5. 5Contextraw.githubusercontent.com ↗

    named, GPT picks the Democrat in 387 races to 37 and Grok declines 441.

    Look up an archived copy ↗

  6. 6Contradicts the claimtechcrunch.com ↗

    when told the names, it refused to pick in 27 out of 30 races

    Look up an archived copy ↗

  7. 7Contexttechcrunch.com ↗

    it picked the Democrat in 24 out of 30 races

    Look up an archived copy ↗

  8. 8Contexttechcrunch.com ↗

    For Grok, it really is the search pattern

    Look up an archived copy ↗

  9. 9Contexttechcrunch.com ↗

    The consumer apps refused to answer, which an OpenAI spokesperson said was significant

    Look up an archived copy ↗

  10. 10Contexttechcrunch.com ↗

    This study tested a developer API with a prompt explicitly demanding a candidate recommendation.

    Look up an archived copy ↗

  11. 11Contexttechcrunch.com ↗

    Parikh noted that, in further testing, he was able to jailbreak the consumer apps with simple prompts

    Look up an archived copy ↗

  12. 12Contexttechcrunch.com ↗

    carried about 1,240 tokens of additional input

    Look up an archived copy ↗

  13. 13Contexttechcrunch.com ↗

    SpaceXAI did not respond in time to comment.

    Look up an archived copy ↗

Members' extras

What each receipt proves, and what would change our mind.

  • What each receipt above proves, in our words
  • The figures table, each number traced to the receipt that states it
  • The trick behind the claim, explained, with the guide to spotting it
  • What would change our verdict
  • Our call on the record, with its due date

The verdict, the opening and every receipt above stay free. Membership adds our analysis to the evidence file for every checked story.

First membership: 30 days free, then A$89 a year. One introductory trial per customer. Card required; renews annually until cancelled. Cancel before the trial ends to avoid the first charge.