Deepfakes bill a good start

Our digital landscape is changing at an unprecedented rate. Generative Artificial Intelligence (genAI) technologies have provided accessible tools for the rapid creation of deepfakes — digitally altered or fabricated content that looks and sounds increasingly realistic. 

There are now estimated to be more than eight million deepfake images on the Internet, growing at a rate of 900 percent each year.

While most of us are likely familiar with the bad, silly and obvious deepfakes fooling some of our less tech-savvy relatives on social media, many are unaware of the more serious harm being caused by deepfakes. 

Deepfake technology isn’t inherently harmful. But with the ability to generate seemingly realistic fictional scenarios, the technology has the potential to cause harm. 

For example, using it to alter a video recording of a political speech, or impersonate a family member to gain someone’s trust (and financial details). This can also extend to virtually undressing people known to the creator, including classmates or acquaintances, or anyone else they have an image of (otherwise known as a form of non-consensual, intimate ‘deepfaking’).

Since X’s AI chatbot Grok allowed people to generate thousands of deepfakes an hour, many of them inappropriate, public awareness of the issue has been increasing. A growing number of people are advocating for regulation of deepfake technologies aimed to protect young people online in particular. 

In our latest Internet Insights survey, 68 percent of those surveyed reported feeling concerned or very concerned about the AI being used to produce harmful content. 

In response to public sentiment, a member’s bill seeking to criminalise non-consensual, explicit deepfakes has recently passed its first reading in Parliament. If made into law, these changes will provide significant protections for people who are harmed by non-consensual, intimate deepfakes.

Why the proposed changes matter 

The Deepfake Digital Harm and Exploitation Bill aims to update relevant laws to criminalise the creation and sharing of non-consensual, ‘intimate’ deepfakes. This would bring New Zealand in line with many of our international peers who have already made similar changes to their laws. 

The proposed protections are critical, as non-consensual intimate deepfakes carry a far higher risk of harm than, say, a funny satire video.

While amounting to only small definition edits to account for ‘synthesised’ images, this legislative shift makes important updates to better account for technological changes that have expanded the types of intimate recordings that can be generated.

It is difficult to estimate the scale or growth of non-consensual, sexually-explicit deepfakes, though Netsafe have observed a ‘surge' in reports of image-based abuse, and other reports of first-hand experience are illuminating the issues caused by lagging legal protections.

This was highlighted in a recent interview of the first New Zealand woman to successfully secure a criminal conviction against the person who had created hundreds of explicit deepfakes using her images. 

Much of the interview focused on Summer’s ordeal pursuing this conviction through legislation that did not clearly criminalise the behaviour and instead relied on a requirement to prove intent to harm (which the article points out is “notoriously hard”).

The changes proposed in this bill aim to remedy this by creating more straightforward legal pathways for people who experience harm, like Summer, removing the need to prove intent in seeking recourse for the damage caused to their reputation and well-being.

What else is needed?

The proposed changes in the Deepfakes bill send a clear message condemning non-consensual intimate deepfakes and take a further step in addressing the harm caused. However, much more is needed to keep Internet users, particularly younger people, safe online.

In our upcoming submission to the Select Committee, InternetNZ is also recommending:

• The establishment of an independent digital regulator responsible for mitigating online harm AND
• That the new independent digital regulator leads a comprehensive digital education programme to support online safety.  

For lawmakers to achieve their aims of online harm reduction, broader changes are required to ensure safety by design and the protection of a free and open Internet. 

This includes ensuring appropriate independent oversight and providing the wider public with the tools and information required to keep themselves and others safe online.

You can see InternetNZ’s submission on the Deepfake Digital Harm and Exploitation Bill here.