If you have gotten a campaign text that fired back a polished reply in seconds, never got snarky, and seemed oddly tireless, there is a good chance you were talking to software, not a staffer. A growing crop of startups is selling conversational bots that handle two-way SMS for campaigns, fielding policy questions and screening potential supporters at scale just as operations ramp up for the 2026 midterms.
Vendors and political operatives told NPR that companies such as Convos and Vector Political are pushing AI agents designed to carry on personalized voter conversations. According to NPR, Vector Political has already sent roughly 2.5 million messages this year, and Convos, which launched in 2025, has signed multiple campaigns. Supporters say these tools help campaigns respond quickly and uncover new persuasion targets. Critics counter that the bots can hide who is actually speaking and risk spreading inaccuracies if a system gets something wrong.
How the Bots Talk Back
The platforms say their agents are trained on campaign-approved scripts, public records, and past messaging so replies stay on brand while generating useful voter data. Tom Carroll, the CEO of Convos, told NPR that the company’s AI agents "will respond within 30 seconds in any language and cut directly to the voter’s question." The companies also say that conversations flagged as sensitive are kicked over to human staffers rather than left to an algorithm.
New Tools Crowd the Field
A widening mix of vendors and agency partners is racing to offer similar firepower. aKillion recently announced a partnership with Subtext to roll out IONCandidate, a conversational product the companies say can juggle thousands of simultaneous text exchanges and funnel the resulting replies straight into voter files, according to a press release cited via GlobeNewswire. Other peer-to-peer texting platforms are layering in AI features of their own and pitching campaign managers on faster, deeper engagement without needing to double the staff roster.
Law, Liability and Trust Issues
Courts, regulators, and wireless carriers have created a patchwork of rules that shapes how these tools can be used and who is on the hook when something goes sideways. The Supreme Court’s 2021 decision in Facebook v. Duguid narrowed the legal definition of an "autodialer," a shift that affects when automated political texts need prior consent, according to the Supreme Court. At the same time, confidence that government can keep up is not exactly soaring. A June 17, 2026 survey found widespread skepticism that regulators can effectively police AI, complicating fights over disclosure requirements and consent rules for political messaging tools, according to Pew Research Center.
What Voters Can Do About It
If you want to cut down on unwanted political outreach, start with the basics. Check any campaign text for the sender’s name and the opt-out line that should be included, then use it. Short replies like "STOP," "END" or "QUIT" are widely recognized by carriers and texting platforms. If something looks shady, you can forward suspicious messages to 7726 ("SPAM") so your wireless carrier can review them, hang on to screenshots, and consider filing complaints with regulators if the texts keep coming. Consumer guidance and legal resources walk through these steps in more detail, including the consumer guide from the FCC and practical how-to advice from LegalClarity.
