regs to riches

Share this post

🪪 deep freaking

www.regs2riches.com

🪪 deep freaking

🏢 policy inaction as a grand scam

Vass Bednar
Feb 19
5
Share this post

🪪 deep freaking

www.regs2riches.com
Twitter avatar for @beccalew
Becca Lewis @beccalew
Much of the internet we know today is not the result of genius tech founders but rather the result of ‘90s moms trading beanie babies on eBay, and I, for one, think that’s nice
1:34 AM ∙ Feb 19, 2023
61Likes1Retweet
Twitter avatar for @coffeebreak_YT
Coffeezilla @coffeebreak_YT
Deepfake scams are here, and we’re not ready.
8:28 PM ∙ Feb 12, 2023
88,865Likes6,805Retweets
Twitter avatar for @motherboard
Motherboard @motherboard
Online trolls created AI voices from video game voice actors and then had them read out victim’s home addresses and posted the results online.
trib.alVideo Game Voice Actors Doxed and Harassed in Targeted AI Voice AttackThe harassment campaign shows that if there is audio of you speaking available online—podcasts, streams, YouTube—you could also be a target of AI-generated voices.
3:40 PM ∙ Feb 19, 2023
18Likes12Retweets

So far, I’ve been pretty good at evading scams. I don’t get my ducts cleaned and I have yet to buy gift cards for my boss. BUT when a deepfake of my boss asks me to buy gift cards for duct cleaning, I’m toast. 

Lately I have found myself feeling curious about deepfakes and scams, as well as deepfakes and advertising. The former should freak us all out as it becomes much harder to discern a Nigerian Prince from your actual husband, and the latter application similarly seeks to leverage our trust to trick us. In both cases, the content is created with the intention to deceive, leaving you  to figure out whether the face and voice seeking to gain your information is legit or pure charlatanism.

Much like the ‘moment’ where we were enamoured with the potential of ‘big data’ before we could anticipate its many misapplications  (i.e. surveillance capitalism), this moment where we are charmed by synthetic media feels similarly blind (**other than deepfake porn, it all seems so…cool?) and even more urgent. 

For instance, the influencer economy is predicated on individuals building digital audiences that they can monetize as a trusted intermediary. We require sponsored content (“sponcon”) to be labelled. But is it still sponcon if the likeness of a person is promoting a product or service? Where would accountability lie in that instance? 

There may be no ‘problem’ with deepfakes in the legal, non-fraudulent sphere. But problems will start when your grandmother gets a deepfake call from her banker, or when the biometric footprint of her voice is mimicked so well that ‘she’ calls her bank and fakes them out. 

Twitter avatar for @emollick
Ethan Mollick @emollick
Creating a credible deepfake of me giving a lecture took only 6 minutes. AIs handled all of it - writing a speech in my style, cloning my voice, and animating an image of me. You really shouldn’t trust any video or audio on the internet ever again.
open.substack.comA quick and sobering guide to cloning yourselfIt took me a few minutes to create a fake me giving a fake lecture.
12:15 AM ∙ Feb 11, 2023
2,697Likes593Retweets

Another issue is that the government and phone companies seem to have lost all control of the cell phone network. Not to seem unpopular, but a lot of my incoming calls seem to be robocall spam linked to fraud centres abroad (almost makes me little more open to that email from the prince). It makes you wonder what the role could or should be for private firms in protecting against deepfake scams, or pledging to not use deepfakes as a tactic to engage in cheaper-than-usual advertising.  

Amidst all this, the big tech platforms are laying off their misinformation teams, which increases the likelihood that we will see or click on junk in our feeds.

Twitter avatar for @nytimesbusiness
NYT Business @nytimesbusiness
Job cuts in the social media industry reflect a trend that threatens to undo many of the safeguards that platforms have put in place to fight disinformation. Many teams assigned to combat false and misleading posts have taken a hit.
nyti.msCombating Disinformation Wanes at Social Media GiantsAs the companies have shed jobs recently, many teams assigned to combat false and misleading information have taken a hit.
6:29 PM ∙ Feb 14, 2023
7Likes5Retweets

So, what happens when the deepfakes get really good? How do you get people to not trust familiar voices on the phone anymore, or even video calls on FaceTime or Zoom?

To date, the government’s approach has generally been to preach vigilance while online firms accept self-attestation that you aren’t a robot during CAPTCHA verification. The CRTC offers guidance to protect yourself against spam. Ontario’s Consumer Protection authority offers education on common scams and how to identify them, and what to do if you’ve been a victim of one. In theory, deepfakes could be considered a form of false and misleading advertising under the federal Competition Act.  

This short take from Torys offers a perspective on deep fakes that is super personal: what if someone deepfakes YOU? It surveys personal privacy, intentional infliction of emotional distress, defamation, breach of confidence and public disclosure of private facts, “false light” (I had never heard of this before!), the impact on businesses, privacy torts, copyright, intentional interference with economic relations and whom to sue. This is all useful if you need to defend your image, but what about when a fake image inflicts harm on you?

Twitter avatar for @satariano
Adam Satariano @satariano
For years, we've heard warning about the use of A.I. in disinformation. Now the first known instance of “deepfake” video tech being used to create fictitious people as part of a state-aligned campaign has been uncovered. w/ @paulmozur
nytimes.comThe People Onscreen Are Fake. The Disinformation Is Real.For the first time, A.I.-generated personas, often used for corporate trainings, were detected in a state-aligned information campaign — opening a new chapter in online manipulation.
11:17 AM ∙ Feb 7, 2023
215Likes155Retweets

The emergent sophistication of deepfakes make it unlikely that recipients will be able to easily recognize a digital racket. The technology demands a stronger policy response: investment that can help us better detect and enforce new standards to combat deepfakes instead of just independently defending our own likeness. 

If we don’t amp up a robust policy response fast, that will be a scam, too. 

Leave a comment

Twitter avatar for @IRPP
IRPP/Policy Options @IRPP
What does it mean for our society when anyone’s voice or image can be manipulated to have them saying or doing anything? @AndrewCichocki says it's time to worry about new breakthroughs in #ArtificialIntelligence. #CdnPoli #CdnTech
options-po.liHow to control AI before it wreaks havocNew technology means it’s time to worry about the effects of artificial intelligence. Steps are needed to restrict access to hardware.
6:01 PM ∙ Feb 3, 2023
Twitter avatar for @zachsilberberg
zach silberberg @zachsilberberg
i spent way too long on this
7:29 PM ∙ Feb 17, 2023
26,805Likes3,911Retweets

⛰️ Anyone reading from Vancouver? I’ll be part of this panel happening on Friday and would love to say hello.

Twitter avatar for @VassB
Vass Bednar @VassB
🔗 Vancouver! Next Thursday evening (Feb. 23rd) we will be chatting at the @internetarcCA about "How to Regulate the Internet." Moderated by @HeidiTworek, I'll be in convo w/ @EmilyLaidlaw, @bhaggart & @christelletono at 5 pm. Free + open to all. eventbrite.ca/e/how-to-regul…
eventbrite.caHow to Regulate the InternetDisinformation, online abuse, privacy breaches, uneven internet infrastructure. How to address these problems confronting Canadians online?
2:48 PM ∙ Feb 14, 2023
16Likes10Retweets

Share this post

🪪 deep freaking

www.regs2riches.com
Previous
Next
Comments
TopNewCommunity

No posts

Ready for more?

© 2023 Vass Bednar
Privacy ∙ Terms ∙ Collection notice
Start WritingGet the app
Substack is the home for great writing