ElevenLabs signs first electoral-authority MoU and expands election-safety safeguards ahead of 2026 votes

ElevenLabs detailed a set of election-focused partnerships and safety measures ahead of a heavy 2026 global election calendar, including its first-ever memorandum of understanding with a national electoral authority.

What's new

ElevenLabs formalized a partnership with Brazil's Superior Electoral Court (TSE) to add a voice layer to the court's 'Electoral Assistant,' letting citizens use voice to get official information through the TSE's website, app, and WhatsApp channels ahead of the country's October elections. The company says it "formalized the partnership through an MoU with the TSE" and joined the TSE's Permanent Program to Combat Electoral Misinformation. It is also training partners at Comprova, a coalition of 42 Brazilian media organizations led by the Brazilian Association of Investigative Journalism, to use its AI Speech Classifier and other detection tools for verification work.

In the United States, ElevenLabs says it is working with policymakers and industry bodies to protect the midterm elections, including membership in California's Countering Tech-Enabled Fraud Task Force to counter voice-enabled scams, a watermarking partnership with Google DeepMind using SynthID, and work with deepfake-detection company Reality Defender.

On the safeguards side, ElevenLabs described its layered defenses against misuse: a "No Go Voices" system that screens voice-clone uploads to block cloning of public officials, political candidates, and other high-risk figures; Voice CAPTCHA, which checks an uploaded sample against a real-time recording of the user reading a generated script; automated real-time and asynchronous content moderation; and a free AI Speech Classifier, in use since 2023, that lets anyone check whether audio was generated on the platform.

Context

The announcement lands as 2026 election activity accelerates worldwide, including major votes in the United States and Brazil, and as voice AI has become expressive and widely available enough to be both a genuine accessibility tool and a credible misinformation risk. ElevenLabs frames the moment directly: "2026 is a significant election year... It's also one of the first election years in which expressive, natural-sounding voice AI is widely available." The company's rivals and partners, including Google DeepMind on watermarking, have faced similar pressure to demonstrate election-safety measures as synthetic voice and video tools proliferate.

Why it matters

A national electoral authority MoU is a meaningfully higher bar than a voluntary safety pledge — it puts ElevenLabs' voice technology directly inside a government's official misinformation-response infrastructure, which is both a trust signal for the company and a template other electoral bodies could adopt. At the same time, the detailed breakdown of technical safeguards (voice-clone screening, CAPTCHA-style liveness checks, watermarking) shows the specific attack surface voice-AI vendors are now expected to defend against by default, not as an afterthought, heading into a dense global election year.