Aspen Digital

Voter suppression has long been a challenge for marginalized communities, but A.I. could exacerbate these issues in this year’s  elections.

Advances in generative AI since the last election means that manufacturing false content at scale and targeting it with precision is faster, cheaper, and more accessible to bad actors than ever before. The risk remains high in the lead up to November and in the immediate aftermath when time and resources for both tech companies and government will be stretched thin.

Participants predicted that AI might be used to target certain communities with disinformation about candidates, or the means and manner of voting. This could come in the form of micro-targeted AI-generated images, videos, in-language message groups, and voice calls.

Based on past experiences with malicious actors targeting communities of color, leaders also voiced concern that deep fakes may be used to spread false but alarming content (e.g., riots at a polling place), or to show election workers appearing to engage in voter fraud.

Concerns about potential AI threats to the democratic process at the local level 

Participants expressed deep concern about the rapid decline of local journalism in many parts of the country, and the rise in fake local news sites. Without “watchdogs” like journalists working to report on and fact-check community-specific matters such as local and state races, mis- and disinformation could spread more rapidly.

There are particular implications for communities where non-English communication is prevalent, as they are often less connected to the broader media ecosystem and are, therefore, prime targets for bad actors looking to spread false information. AI-generated translation makes this form of manipulation faster and cheaper.  

Challenges driven by AI have the potential to be worse at the local and state level because of decreased public engagement in down-ballot races This may be exacerbated during non-presidential years, when public awareness of elections and candidates is more limited 

Communication across sectors must be stronger.

Leaders in every sector asked for more coordination among communities, election officials, and tech companies to help them mitigate potential threats to their democratic and civil rights. 

Community leaders expressed a strong belief that tech companies have a responsibility to create clear and useful processes for engagement with community leaders and election officials. They expressed that the process of engaging with tech companies is currently too opaque and lacks consistency within individual companies and across the industry

Community leaders are also calling on technology companies to better prepare their communities with information about how to recognize targeted AI-driven mis- and disinformation and how best to report it.

Tech executives  agreed with community and election leaders on the need to streamline communication. Oftentimes tech leaders will give community representatives  their direct contact information to help cut through the ambiguous process but acknowledge that this is not a scalable solution.

Lack of representation leads to flawed innovation and increasing distrust.

For the long-term, participants emphasized the need for innovation to be “done by us” not “done to us,” especially when it comes to designing and testing new products. 

Leaders across the board emphasized how targeted engagement with communities can help build trust in the voting process, especially when prefaced by an acknowledgment of past harms. Local members of the community are best positioned to educate technology companies on the cultural specificities of under-represented groups in order to better equip platforms to address hyper-targeted attacks. 

Participants stressed that ethical guardrails for AI products are not enough to predict or mitigate harm to a community if they are not also represented in the design or testing stages. Given the recent deep cuts to DEI commitments across the industry, this issue has become particularly acute. Community leaders discussed the idea of “bridging” roles within technology companies – product development experts from underserved communities who have first-hand knowledge about how their specific communities interact with tech products.   

There was also commentary that the suppression of democratic rights is not limited solely to elections, The group shared that without ongoing engagement by tech companies, AI-driven harms against communities will only continue to snowball.