New York Attorney General Letitia James on Wednesday, Oct. 13, released an updated guide to help New Yorkers spot, avoid and report election misinformation created by artificial intelligence (AI) ahead of the November 3 general elections. The new guide, issued by the Office of the Attorney General (OAG), updates and expands earlier guidance to reflect a landscape in which AI-generated images, videos, and audio recordings have grown dramatically more sophisticated. James said that, in just the last year, AI has produced content so realistic that ordinary voters can no longer rely on their instincts to tell the difference between a genuine recording and a digital forgery. The OAG has also documented instances in which AI chatbots, the automated programs that answer questions on websites and apps, provided voters with incorrect or outdated information about poll site locations, voting deadlines, and voting requirements. The attorney general said the guide offers practical steps New Yorkers can take to check election information against official sources before they act on it or share it with others. “When New Yorkers head to the polls, they deserve to have accurate information to cast an informed vote,” James said. “AI has made it easier than ever for bad actors to spread convincing lies about candidates and about when, where, and how to vote, and chatbots can confidently give voters the wrong answer.” She added that her office would be “standing guard throughout this election” and would not allow anyone to use AI to deceive voters or keep New Yorkers from the polls. With early voting scheduled to begin on Oct. 24 and Election Day falling on Nov. 3, the guide is intended to serve as a practical resource for voters navigating what is expected to be a high-intensity information environment over the next three weeks.
AI tools can create photos, videos, and audio recordings that appear to show real people saying things they never said or doing things they never did, according to James. They can also generate fake images, messages, letters, websites, and social media posts designed to look like official government notices or official election communications. This is not a hypothetical concern. The attorney general said bad actors can use this fake content to mislead voters about candidates, voter eligibility, and where, when, or how to vote. A fake video might show a government official giving false voting instructions, such as telling voters to return their absentee ballot to the wrong address or to bring a specific document that is not actually required. A fake text message or social media post might try to scare voters into staying home by threatening a consequence that would never happen, such as arrest for unpaid parking tickets at the polls or automatic deportation for appearing on the voter rolls. A robocall might use AI-cloned audio that sounds like a campaign volunteer, a local election clerk, or even a candidate, urging voters to show up on the wrong day or to “confirm” their eligibility by providing a Social Security number. Because these materials can be generated in seconds and distributed across multiple platforms at low cost, they can be targeted at specific neighborhoods, demographic groups, or partisan communities. Once a convincing fake is shared by a trusted friend or family member, it gains credibility and can spread quickly through private messaging groups, where election officials cannot easily monitor or correct it. The OAG guide is designed to stop that chain of transmission by making voters more skeptical and more likely to verify before forwarding.
The updated guidance reflects how quickly the technology has evolved since the attorney general first released it. The original guide told voters to watch for warning signs like unnatural body movements, odd lighting, strange shadows, or distorted sound in videos. James said those indicators are no longer reliable. “Today’s AI fakes are far more convincing, and those warning signs may no longer be present,” she said. Modern generative models can create lifelike images and high-quality audio that mimic the cadence, accent, and appearance of a real person with remarkable accuracy. Some deepfakes have fooled experts, and the pace of improvement shows no signs of slowing. The OAG also cautioned that tools claiming to detect AI-generated content are not always accurate. Automated detectors can be fooled by compression, downscaling, or overlays; they can produce false confidence when they flag authentic content as fake or miss sophisticated forgeries entirely. Social media companies have developed labeling systems for synthetic media, but those labels are inconsistent and can be stripped when content is re-uploaded or cross-posted. The guide therefore tells voters not to rely on visual inspection or automated detection. Instead, it asks them to take a completely different approach: before believing anything surprising about the election, check the information with a trusted, official source. That might mean looking at the website of the New York State Board of Elections, calling a local county board of elections, or confirming a report through an established news organization. By making verification a habit, voters can protect themselves even if they cannot identify a deepfake with certainty.
Under New York law, any political communication that includes seemingly realistic but AI-generated or otherwise deceptive images, video, or audio must be clearly labeled as manipulated. This requirement applies to campaign advertisements, digital mailers, and other communications distributed within the state. The updated guide explains the labeling rules, including what counts as a qualifying disclosure, where it must be placed, and who is responsible for ensuring that it appears. It also notes that the law is intended to give voters the chance to know when they are looking at synthetic content rather than a genuine recording of a candidate or government official. However, the OAG cautioned that a great deal of fake content posted on social media, messaging apps, and websites does not carry any label. Foreign operations, anonymous accounts, and independent actors may ignore the law, and platforms may not catch every violation. Tools that claim to detect AI-generated content are not always accurate, the OAG said, so the absence of a label is not evidence that the content is authentic. Attorney General James encouraged anyone who encounters a voting-related video, image, or audio recording that they suspect to be manipulated and that does not include a disclosure to file a complaint with her office. The complaint process is straightforward: voters should provide a link or screenshot of the content, the name of the platform where they saw it, the account or person who shared it, and any additional context. Each complaint helps the OAG spot patterns of coordinated disinformation and build cases against the individuals and organizations responsible. James said her office will investigate violations of the state’s election law and use all available enforcement tools to stop deceptive AI-driven communications.
In addition to fighting deepfakes, the OAG is warning New Yorkers about the dangers of AI chatbots. Chatbots are software tools that use large language models to answer questions in natural-sounding text. They are increasingly integrated into search engines, social media apps, customer service portals, and dedicated election-information websites. But these systems are not reliable sources for official election information. The attorney general said the OAG found that chatbots can give voters information for the wrong city or state, incorrect deadlines, and false voting requirements. A voter might ask, “Where do I vote in Brooklyn?” and receive the address of a polling place in Buffalo or a building that closed years ago. Another might ask, “Can I register to vote on Election Day?” and receive a confident “yes” even though New York’s registration deadline has passed or does not permit same-day registration in that jurisdiction. Some chatbots have even invented requirements that do not exist, such as a demand for proof of citizenship or a notarized affidavit to vote by mail. These errors happen because chatbots generate answers based on patterns in their training data, not from a verified database of current election law. They may sound polite and convincing, but they can be completely wrong. “Voters who rely on a chatbot without checking an official source could end up at the wrong poll site or miss an important deadline,” James said. The guide urges voters to treat chatbot responses as a prompt for further research, not as a final answer. For accurate voting information, James recommends contacting the New York State Board of Elections or a local board of elections. Voters can use the state board’s county board roster tool to locate their county election office and the voter lookup tool to confirm their registration, find their assigned poll site, and check their mail-in or early voting ballot status.
With early voting running from Saturday, Oct. 24 through Sunday, Nov. 1, and Election Day on Tuesday, Nov. 3, the next three weeks are the most vulnerable period for AI-fueled misinformation. James said the updated guide gives New Yorkers practical steps to check election information with official sources before they act on it or share it. She encouraged voters to be especially careful when they come across content that provokes anger, fear, or excitement, as those emotions are precisely what manipulators exploit. The guide also recommends discussing AI misinformation with family and friends, particularly those who may be less familiar with how realistic deepfakes have become. The attorney general noted that older residents, new voters, and non-native English speakers can be disproportionately affected by convincing AI-generated robocalls, texts, and videos. She asked New Yorkers to help each other by sharing official voting resources and by reporting suspected manipulated content rather than retweeting it. “Before you believe or share anything you see about the election, check it with your board of elections or another trusted source,” she said. The OAG has made the guide available on its website and has established mechanisms for voters to file complaints about election misinformation, including AI-manipulated media that appears without the required disclosure. James said her office will monitor the information environment throughout the election season and will take aggressive action against anyone using AI to deceive, intimidate, or mislead voters. “We will not let anyone use AI to deceive voters or keep New Yorkers from the polls,” she said. In a state with millions of registered voters, even a small amount of targeted misinformation can have an outsized impact, which is why the attorney general is urging all New Yorkers to adopt a simple habit: verify before you trust, and trust official sources.