Disinformation security, part 2: fighting fire with algorithmic fire

Intro

In Part I, we explored how disinformation can tank stock prices, shred trust, and turn CEOs into unwitting deepfake villains. Now, let’s arm your brand with the tools to fight back. Spoiler: It’s not just about debunking lies—it’s about out-innovating the innovators of chaos.

Core areas to watch

1. 1

   Human writer or AI? Scholars build a detection tool

   SourceStanford HAI

   Executive summary Stanford researchers have developed DetectGPT, a tool capable of identifying AI-generated text with up to 95% accuracy by analyzing linguistic patterns and anomalies. This advancement addresses growing concerns about the misuse of AI in areas such as education and social media. DetectGPT offers a means to discern between human and machine-authored content, contributing to the broader effort to maintain authenticity in digital communication.

   Key takeaways

   • DetectGPT employs a zero-shot learning approach, eliminating the need for training on labeled datasets or watermarking to identify AI-generated text.
   • DetectGPT enhances the ability to maintain content authenticity and integrity in various digital communications.
   • The tool is designed to assist educators and institutions in detecting AI-generated content, addressing concerns about academic integrity in the age of advanced language models.

   Hot takeIf you’re not using AI to fight AI, you’re bringing a knife to a drone fight.

2. 2

   A Russian Bot Farm Used AI to Lie to Americans. What Now?

   SourceCSIS

   Executive summary In 2024, U.S. authorities dismantled a Russian AI-powered bot farm that disseminated pro-Kremlin disinformation through nearly 1,000 fake social media accounts impersonating Americans. This operation, developed by Russia's state-run RT and managed by the FSB, utilized AI tools like Meliorator to create convincing personas and spread propaganda across platforms such as X (formerly Twitter). The incident underscores the growing threat of AI-enhanced disinformation campaigns and the need for robust countermeasures.

   Key takeaways

   • First-of-its-kind: This was the first known disruption of a Russian-sponsored, AI-enhanced social media bot farm by international authorities.
   • 968 Number of fake social media accounts seized by U.S. authorities, all created using AI to impersonate Americans.
   • The disinformation targeted audiences in the United States, Poland, Germany, the Netherlands, Spain, Ukraine, and Israel.

3. 3

   Most Consumers Can’t Identify AI-Generated Fakes

   SourceiProov

   Executive summary A recent study by iProov reveals a significant human vulnerability to AI-generated deepfakes, with only 0.1% of participants accurately identifying all fake content. The research highlights widespread overconfidence in detection abilities and a general lack of awareness, especially among older adults. These findings underscore the urgent need for advanced technological solutions to combat the growing threat of deepfake-driven misinformation and fraud.

   Key takeaways

   • 0.1% of participants correctly identified all deepfake and real stimuli, even when primed to detect fakes.
   • 22% of consumers had never heard of deepfakes prior to the study, indicating a significant awareness gap.
   • 74% of respondents expressed concern about the societal impact of deepfakes, with 68% specifically worried about misinformation.

4. 4

   US House of Representatives call for legal liability on Deepfakes

   SourceFuture of Life Institute

   Executive summary The U.S. House of Representatives is advocating for legal accountability across the entire deepfake supply chain, targeting developers, distributors, and users of AI-generated synthetic media. This initiative, supported by a coalition including the Future of Life Institute, SAG-AFTRA, and Equality Now, aims to mitigate the escalating threats posed by deepfakes to democracy, personal safety, and public trust. Proposed measures encompass civil and criminal liabilities, mandatory content removal, and obligations for AI and cloud service providers to prevent the misuse of their technologies.

   Key takeaways

   • Between 2022 and 2023, deepfake-related fraud incidents surged by 3,000%, highlighting a significant rise in AI-driven scams.
   • The TAKE IT DOWN Act, passed in April 2025, mandates that online platforms remove reported non-consensual deepfake content within 48 hours, aiming to curb the spread of harmful synthetic media.
   • The NO FAKES Act, introduced in May 2025, seeks to hold individuals and companies accountable for unauthorized digital replicas, with nearly 400 artists endorsing the bill.

5. 5

   AI misinformation: Here's how to reduce your company’s exposure and risk

   SourceIBM

   Executive summary The article discusses the challenges posed by AI-generated misinformation and offers strategies for organizations to mitigate associated risks. It highlights how AI models, particularly generative AI, can inadvertently produce false or misleading content due to their predictive nature and phenomena like AI hallucinations. The article emphasizes the importance of implementing safeguards, such as ensuring data quality and deploying retrieval-augmented generation, to reduce the likelihood of AI systems generating misinformation.

   Key takeaways

   • Social media platforms like TikTok and Facebook have started labeling AI-generated content to combat the spread of misinformation.
   • 53% of Americans rely on social media for news, increasing their exposure to AI-generated misinformation.
   • 12.5 billion dollars In 2024, Americans lost this amount to phishing attacks and other fraud, a figure expected to rise with the increased use of generative AI by scammers.

Case study spotlight

The wrap-up

Disinformation isn’t just a risk—it’s a Darwinian test. Brands that adapt will survive; those that don’t will become cautionary tales. The tools exist, but victory requires speed, alliances, and a dash of paranoia.

Call to action

1. Simulate. Run a deepfake attack drill with your executive team this quarter.
2. Collaborate. Join industry coalitions like the Partnership on AI or Trusted Web Initiative.
3. Debate. Join the discussion on LinkedIn: “Should governments license AI tools to curb disinformation?”

Final noteIf your CFO still thinks “blockchain” is a gym machine, forward this. They’ll thank you later.

Your turn