Ethical Considerations in AI Expansion

A new generative AI advancement is introduced to the world?through a research paper or carefully worded blog post.

Thought leaders and notable names gush over how it will revolutionize the world and upend?something?(for good, perhaps). When it is slowly rolled out to exclusive partners, its jaw-dropping qualities are treated as a spectacle, shared countless times over social media and on the news. After the public finally gets a hold of it, users find innovative new ways to implement it and collectively blow everyone's minds.?

At the same time, bad actors have concurrently figured out how to use this technology?to abuse or defraud people en masse. The parent company stays noticeably silent or, worse, talks about non-specific actions they will take to one day combat abuse. (Maybe.)

Today on Reality Defender, I dive into the current cycle of generative AI release and response — how it accelerates abuse, fraud, and disinformation — and what we can do to change it.

Innovation and responsibility can exist in tandem, and by disrupting the current way of building without considering the consequences, we can have both the most advanced AI systems?and avoid a litany of harms they would otherwise bring.

- Ben Colman , Co-Founder/CEO, Reality Defender

A Brief History of Deepfakes

Deepfakes originated in the 1990s and gained traction in the 2010s, with major advances in the field in 2014.

• Learn the history of deepfakes' technological origin.
• Hear about key breakthroughs as it advanced.
• See how deepfake technologies continue to advance and influence our world.

Reality Defender at RSA

Reality Defender?is proud to be part of this year's RSAC Innovation Sandbox.?Learn more about this amazing opportunity here, and join us in SF next month for the RSA Conference.

News

Deepfakes

• Just because the Kate Middleton video has an editor's note does not mean it's a deepfake.?
• OpenAI's new voice model is too risky for general release — but platforms are using it with permission anyway.

Policy and Government

• A new AI pact between the U.S. and the U.K. includes a cooperation on sharing AI safety and model information.
• New Hampshire passed AI election rules months after a deepfake robocall of President Biden was unleashed on citizens.
• No, that's not really Justin Trudeau on YouTube selling trading apps.
• You won't find Microsoft Copilot on Congress-owned devices anymore.
• Every U.S. agency may be getting a chief AI officer.

AI

• Apple's new LLM may be able to "beat" GPT-4.
• How does generative AI view queer people? Wired takes a deep dive into perceptions and biases within models.?
• Peer reviews of research papers may be no match for generative text.
• More people are using generative AI to do their work.
• OpenAI and Microsoft may be building a $100 Billion AI supercomputer.

Arts and Entertainment

• AI is coming to the Xbox.
• Hollywood execs may be getting in on the Sora hype.
• The WGA is urging for legal protections of writers from AI.

\ Case Study

How Reality Defender Exposed a Political Disinformation Campaign in Taiwan

A sophisticated deepfake campaign targeted Taiwan's presidential election. Reality Defender partnered with authorities to rapidly detect the fake audio and protect democracy.

Thank you for reading the Reality Defender Newsletter. If you have any questions about Reality Defender, or if you would like to see anything in future issues, please reach out to us here.