WASHINGTON – U.S. Sens. Mark R. Warner (D-VA), Vice Chairman of the Senate Select Committee on Intelligence, Alex Padilla (D-CA), Ranking Member of the Senate Committee on Rules and Administration, and Representatives Jim Himes (CT-04), Ranking Member of the House Permanent Select Committee on Intelligence, and Joe Morelle (NY-25), Ranking Member of the House Administration Committee, urged leading technology companies to ensure their platforms and tools help protect the integrity of our elections and support free, fair, and secure midterms.



This letter follows up on Sen. Warner’s letter from March where he pushed tech companies to take action against deepfakes and maliciously manipulated media. Additionally, this letter references the 2024 Tech Accord to Combat Deceptive Use of AI, where leading technology companies – including some of those receiving this letter – pledged to work together to detect and counter harmful AI content surrounding the 2024 elections.



In a letter to the CEOs of Adobe, Alphabet, Anthropic, Black Forest Labs, Bluesky, Discord, ElevenLabs, LinkedIn, Meta, Microsoft, Midjourney, Mistral, OpenAI, Perplexity, Pinterest, Reddit, Reflection AI, Rumble, Snap, SpaceX, Stability AI, Telegram, Thinking Machines Labs, TikTok U.S., and Truth Social, the legislators wrote, “Your platforms play a significant and important role in sharing, disseminating, and in some cases generating information about those elections. Much of the United States—and the world—searches for and receives information from American search engines, American chatbots, and American social media platforms. These connections create opportunities for Americans to come together and debate important topics and share information. But they also create vectors for foreign adversaries and domestic actors to easily and quickly reach—and manipulate—large audiences.”



The legislators explained that their social media platforms and AI tools can quickly generate, execute and disseminate influence campaigns that used to take hundreds of hours of labor. As Americans increasingly turn to AI tools to inform themselves about political candidates, ballot issues, and the logistics of voting, AI tools could be used to spread inaccurate, outdated, manipulated, or biased information.



The legislators reminded the CEOs that many of their companies have committed to ensuring they were not complicit in harming elections or hampering the ability of Americans to vote. Additionally, while many have also committed to authenticating and labeling digital media and adopted policies to increase the odds that users see the most reliable voting and election information, there have been demonstrated inconsistencies among the companies regarding the effectiveness and applicability of their respective policies.



“Unfortunately, there are concerning signs of backsliding on some of your commitments to protecting elections from misuse of your products. These signs are particularly alarming because they come at a time when the capabilities of malign actors are growing in scale and sophistication and our elections are under attack from multiple directions… Public reporting indicates that Russia continues to undermine confidence in U.S. elections through disinformation campaigns executed directly on many of your platforms; though such efforts clearly violate your terms of service, some of your companies have not quickly or consistently removed such content,” wrote the legislators.



They continued, “Equally alarming is that some of your AI chatbots have an unacceptably high error rate when responding to queries about election and voting processes, subjecting users to confusing or inaccurate answers that could result in voter disenfranchisement. And some of your content removal and fact-checking policies are woefully insufficient or even nonexistent, allowing false, manipulated, and misleading content to continue to circulate on your platforms unchecked. In the past, the Federal Trade Commission has notably pursued enforcement actions against companies that don’t live up to the self-regulatory commitments they make, including in a 2012 case against Google for failing to follow the National Advertising Initiative code of conduct.”



The legislators then recommended a list of concrete actions the companies could take, including removing all demonstrably false information related to election and voting logistics once they find it, using independent fact checkers, regularly updating automated tools to review and remove or reduce the reach of false election-related content, and directing all users to authoritative election-related resources.



In regards to generative AI, the legislators recommended actions such as making sure AI-generated responses to questions about election and voting logistics source their information from state and local election offices and trusted news organizations, incorporating disclaimers and flags in AI-generated responses if election and voting logistics may be outdated, and making sure that all AI-generated images and videos are labeled and can be identified as AI-generated content both by users and by technology-based detection tools.



“Again, we strongly urge you to enhance—not reduce—your election security efforts in the final days before the 2026 midterms and to maintain strong efforts in this area on an ongoing basis ahead of future elections as well,” the legislators concluded.



The full letter is available here and below.



We write to urge you to adopt policies and programs that ensure your platforms contribute to, and do not detract from, the effort to ensure that America’s elections are free, fair, and secure.



As the CEOs of America’s largest technology companies, you are aware of the great responsibility that your companies hold in America’s elections. Your platforms play a significant and important role in sharing, disseminating, and in some cases generating information about those elections. Much of the United States—and the world—searches for and receives information from American search engines, American chatbots, and American social media platforms. These connections create opportunities for Americans to come together and debate important topics and share information. But they also create vectors for foreign adversaries and domestic actors to easily and quickly reach—and manipulate—large audiences.



Moreover, American companies’ AI offerings offer both substantial benefits and substantial risks. Your tools can quickly generate, execute, and disseminate entire influence campaigns. While it previously would take hundreds of manhours to achieve such effects, your tools speed and streamline efforts to create material that are inaccurate or false, but nevertheless convincing. Increasingly, some American voters turn to AI tools to research candidates and ballot issues and seek information about voting logistics, increasing the risk of exposure to inaccuracies, outdated or manipulated information, or biased summaries.



Many of your companies have demonstrated a commitment over the past decade to ensuring that your platforms are not used to harm elections or hamper the ability of Americans to vote. After the 2016 elections, many of your companies turned particular focus to identifying and eliminating coordinated inauthentic behavior on your platforms and establishing searchable archives for election ads. With the rise of powerful generative AI tools, your companies further committed to authenticating and labeling digital media, including by joining the Coalition for Content Provenance & Authenticity (C2PA) and signing the Tech Accord to Combat Deceptive Use of AI in 2024 Elections. Many of your companies have also adopted policies seeking to increase the odds that your users are exposed to only the most authoritative and reliable voting and election information. We believe maintaining and enhancing these kinds of commitments is more important than ever heading into the final days of the 2026 midterm election cycle.



Unfortunately, there are concerning signs of backsliding on some of your commitments to protecting elections from misuse of your products. These signs are particularly alarming because they come at a time when the capabilities of malign actors are growing in scale and sophistication and our elections are under attack from multiple directions. Despite many of your companies having policies prohibiting deception and requiring the labeling of AI-generated content, there have been demonstrated inconsistencies regarding the effectiveness and applicability of your policies. In many cases, some of your social media platforms and AI tools still allow for the creation and/or dissemination of false, synthetic, misleading, or inauthentic election content in direct violation of your policies. Public reporting indicates that Russia continues to undermine confidence in U.S. elections through disinformation campaigns executed directly on many of your platforms; though such efforts clearly violate your terms of service, some of your companies have not quickly or consistently removed such content.



Equally alarming is that some of your AI chatbots have an unacceptably high error rate when responding to queries about election and voting processes,  subjecting users to confusing or inaccurate answers that could result in voter disenfranchisement. And some of your content removal and fact-checking policies are woefully insufficient or even nonexistent, allowing false, manipulated, and misleading content to continue to circulate on your platforms unchecked. In the past, the Federal Trade Commission has notably pursued enforcement actions against companies that don’t live up to the self-regulatory commitments they make, including in a 2012 case against Google for failing to follow the National Advertising Initiative code of conduct.



Information related to election and voting logistics—a category that includes registration deadlines, polling hours and locations, requirements about voter ID, rules for mail or provisional voting and ballots, and voter eligibility—is information that is integral to the functioning of our democracy. When objectively false information about election administration is introduced and disseminated with the intent to deceive or to discourage prospective voters, your companies must have robust and consistently applied rules to differentiate that misinformation from ordinary political speech about elections and to take steps to remove that misinformation.



We continue to strongly encourage you to take the actions and adopt the best practices outlined in Senator Warner’s March 2026 letter on anticipating, identifying, and countering malicious manipulated media. These best practices are important steps to protecting vulnerable communities, public trust, and democratic institutions.



We also encourage you to adopt specific policies and actions regarding election-related content. To the extent that your company offers social media or other content access or distribution platforms, we strongly encourage you to take the following actions to keep your platforms free from coordinated inauthentic behavior and election-related misinformation and disinformation:



Immediately remove all demonstrably false information related to election and voting logistics upon becoming aware of it—including information about where and how to register to vote, and where and how to vote.
Support dedicated reporting tools on your platforms for election officials to report examples of false election information.
Improve the speed by which your tools detect and remove false information related to voting logistics and political and election-related violence.
When such tools are used to detect false, misleading, and inauthentic content, ensure that the tools are kept updated, and that their findings are promptly subject to human review and action to remove violative content.
Utilize independent fact checkers in addition to internal fact-checking teams and regularly updated automated tools to review and remove, or at least reduce the reach, impressions, and discoverability of, false, misleading, or inauthentic election-related content.
Engage independent media and civil society organizations to assist in efforts to verify media, generate authenticated media, and educate the public.
Notify users when they have engaged with election-related content that has subsequently been determined to be false, misleading, or inauthentic.
Affirmatively direct all users to authoritative election-related resources, especially if users were determined to have engaged with false, misleading, or inauthentic election-related content.
Immediately remove all content advocating for violence of any kind upon becoming aware of it, but particularly content advocating political and/or election-related violence.
Publish regular reports detailing efforts to identify and remove false election-related information, election-related misinformation and disinformation, election and political violence, and coordinated inauthentic behavior.


To the extent your company offers generative AI products, we strongly encourage you to take the following actions to keep your AI systems and tools safe as they engage with and generate election-related information:



Ensure that AI-generated responses to questions about election and voting logistics pull only from highly authoritative sources and do not pull from social media platforms or other user-generated content. Authoritative resources should be limited to state and local election offices and top trusted news organizations, as well as civil society organizations aggregating content from these other authoritative and reliable sources.
Provide additional resources to pre-deployment guardrail testing for AI-generated election-related content, as well as ongoing reviewing of model performance.
Strengthen guardrails to prevent AI systems from responding to queries with material that can be used in election disinformation campaigns.
Implement continuous auditing until the November federal elections, and for an appropriate time as ballot counting, litigation, or challenges to the results remain ongoing, for accuracy and objectiveness in AI-generated responses to election-related questions, including mechanisms by which users can report inaccurate answers.
Incorporate disclaimers and flags in AI-generated responses if election and voting logistics may be outdated, and direct users to authoritative resources.
Provide inline citations and links for all election-related information provided in an AI-generated response.
Encourage users to independently verify AI-generated answers to questions about election and voting by reviewing authoritative and reliable resources.
Ensure that all AI-generated images and videos are consistently clearly labeled and can be identified as AI-generated content both by the human eye and by technology-based detection tools.
Ensure that election-related responses provided in languages other than English are accurate and reliable.


Many of you have announced content partnerships with highly reliable sources to ensure that your AI and social media platforms only surface the most accurate and reliable information related to election and voting logistics and election results. We strongly support this practice and encourage you to ensure that your platforms properly identify and curate the authoritative and reliable resources from which they pull from on election-related topics to avoid sharing outdated, inaccurate, unreliable, or otherwise incorrect information on this critical topic area that you cannot afford to get wrong. In addition, we urge you to be mindful of Spanish and other foreign language content, which plays a large and important role in the U.S. electorate but often receives less attention and response than online English-language election misinformation.



We also encourage you to be transparent about the efforts that your companies are making on these fronts. Public reporting indicates that in recent years, several of your companies have conducted layoffs, including of personnel on the trust and safety teams responsible for ensuring your platforms’ integrity when it comes to election topics. Firing such personnel after there has been demonstrated evidence and indicators of orchestrated and coordinate foreign misinformation operations exposes your platform to various integrity threats and is cause for serious concern.



Particularly as the 2026 midterm elections are officially underway, your companies should be increasing the staffing and capabilities of your respective trust and safety teams, especially those teams responsible for avoiding platform exploitation relating to election topics. Refusing to grow and expand your election integrity teams—and even shrinking these teams—does not support your stated commitments to countering harmful election content and suggests to bad actors that your policies are not enforceable.



Finally, third-party researchers play a critical role in studying and producing independent reviews of social media platforms and AI systems and tools, particularly as it pertains to auditing the effectiveness of your trust and safety policies. Your companies should collaborate with third-party researchers by granting them access to your products and data, including data on how users are engaging with your platforms on election-related issues, and allow for the testing of your platforms and AI tools.



Again, we strongly urge you to enhance—not reduce—your election security efforts in the final days before the 2026 midterms and to maintain strong efforts in this area on an ongoing basis ahead of future elections as well. We thank you for your attention to these important measures and we look forward to hearing about your efforts in these areas.