AI Deepfake Election: The upcoming 2024 White House race is bracing itself for a surge in AI-enabled disinformation, with a recent robocall impersonating President Joe Biden sounding alarms about the potential misuse of audio AI deepfakes. The robocall, employing Biden’s voice, urged New Hampshire residents not to vote in the Democratic primary, leading to an investigation into possible voter suppression. As AI-powered applications become more accessible, researchers anticipate an increase in AI deepfakes during this election cycle, raising concerns about their impact on electoral integrity.
The disinformation landscape is evolving with AI tools, particularly voice cloning, becoming widespread. Pindrop, a cybersecurity firm, revealed that a text-to-speech system from ElevenLabs was used to create the Biden robocall, underlining the increasing sophistication of AI manipulation. Campaigners and tech investors are leveraging advanced AI tools for effective political messaging, with voice cloning startups attracting significant investments.
This incident intensifies the debate surrounding AI-generated robocalls, prompting discussions about potential regulations to address the issue. The call for safeguards around generative AI tools or even a ban on robocalls is gaining momentum. Policymakers are under pressure to implement protective measures to prevent potential electoral chaos and preserve trust in the democratic process.
The political deepfake landscape is at a critical juncture, and the fake Biden call serves as a stark reminder of the challenges associated with AI manipulation. As regulators consider the legality of AI-generated robocalls, the need for comprehensive protections against deepfakes, especially in the audio domain, has become more urgent.
The proliferation of AI audio tools, outpacing detection capabilities, demands innovative solutions. Recommendations include incorporating audio watermarks or digital signatures into tools and implementing regulations that restrict access to verified users. Ensuring election integrity becomes a priority, requiring investment in trust and safety measures to combat the evolving threat of AI-enabled disinformation.
AI Deepfakes: Understanding, Creation, and Legal Implications
What is an AI Deepfake? An AI deepfake refers to a synthetic media, typically audio or video, created using artificial intelligence (AI) techniques to alter or replace the likeness of a person in a way that appears authentic. The term “deepfake” is a combination of “deep learning” and “fake,” reflecting the use of deep neural networks to generate highly convincing and often deceptive content.
How is it Made?
- Data Collection: Deepfakes start with gathering extensive datasets of the target person’s images or audio recordings.
- Training the Model: AI algorithms, particularly deep neural networks, are trained on these datasets to learn and mimic the subtle nuances of the person’s facial expressions, voice, or mannerisms.
- Generating Content: Once trained, the model can generate new content by combining elements from different sources or altering existing content, creating a realistic-looking and sounding output.
Laws Governing Deepfakes: The legal landscape around deepfakes is evolving, and regulations may vary across jurisdictions. Key legal considerations include:
- Fraud and Misrepresentation: Deepfakes used for fraudulent purposes, such as impersonation or creating fake content with malicious intent, may violate laws related to fraud, identity theft, or misrepresentation.
- Privacy Laws: Unauthorized use of someone’s likeness for creating deepfakes may infringe on privacy rights. Jurisdictions with robust privacy laws may provide individuals with legal remedies against such unauthorized use.
- Defamation: Deepfakes that harm an individual’s reputation or spread false information may be subject to defamation laws. Legal action can be taken if the content damages a person’s character or reputation.
- Intellectual Property: Using deepfakes to replicate copyrighted material without permission may violate intellectual property laws. Rights holders can pursue legal action for unauthorized use of their content.
- Cybersecurity and Unauthorized Access: Unauthorized access to personal data or systems to create deepfakes may breach cybersecurity laws, leading to legal consequences.
- Election Laws: In the context of political deepfakes, laws governing elections may address the spread of false information, influencing voters, or engaging in deceptive practices.
- Consent and Right of Publicity: Depending on jurisdiction, laws may require obtaining consent from individuals before using their likeness for commercial purposes or publicizing deepfakes.
As the threat of deepfake technology grows, policymakers are working to adapt existing laws and create new regulations to address the challenges posed by synthetic media. It’s crucial to stay informed about local and international legal frameworks to navigate the complex landscape of AI deepfakes responsibly.
Several AI deepfake scandals have gained attention in the past, highlighting the potential misuse of synthetic media for various purposes, including misinformation, fraud, and privacy breaches. Here are some notable instances:
- Deepfake Pornography:
- Date: 2017 onwards
- Details: Deepfake technology has been frequently used to create explicit fake videos featuring celebrities’ faces superimposed onto adult content performers. Many of these videos have been shared without the subjects’ consent, leading to privacy concerns.
- Jordan Peele’s Barack Obama Deepfake:
- Date: 2018
- Details: Filmmaker Jordan Peele collaborated with Buzzfeed to create a deepfake video featuring former President Barack Obama delivering a PSA about the dangers of deepfakes. The video aimed to raise awareness about the technology’s potential for misinformation.
- Mark Zuckerberg’s Deepfake:
- Date: 2019
- Details: An AI-generated video featuring Facebook CEO Mark Zuckerberg was created by artists Bill Posters and Daniel Howe. The video, intended as an art installation, showcased the technology’s ability to manipulate high-profile figures for artistic and political commentary.
- Deepfake Voice Scam:
- Date: 2019
- Details: Criminals used deepfake voice technology to mimic the voice of a company executive and instruct an employee to transfer funds. The scam highlighted the potential for deepfakes to facilitate financial fraud through voice manipulation.
- Political Deepfakes in Belgium:
- Date: 2019
- Details: Belgian political party Socialistische Partij Anders (sp.a) created deepfake videos featuring leaders of other parties seemingly expressing support for their political opponent. The stunt aimed to draw attention to the issue of misinformation in politics.
- Tom Cruise Deepfake on TikTok:
- Date: 2021
- Details: Deepfake videos featuring an uncanny impersonation of Tom Cruise circulated on TikTok, raising concerns about the technology’s potential for impersonation and deceptive practices.
These scandals underscore the ethical, privacy, and security challenges posed by AI deepfakes. The incidents have prompted discussions about the need for legal frameworks, awareness campaigns, and technological solutions to address the risks associated with the widespread use of synthetic media.
Shayne Heffernan
