ChatGPT, Bard, Claude Tricked to Generate Malicious Content

We’re yet to see political deepfakes that are convincing enough to spark wars or threaten democracy in a meaningful way. But the technology is only going to become more powerful, accessible and convincing, and right now, the people making deepfakes for nefarious means have little reason to stop. These laws may be unable to prevent bad actors from making non-consensual deepfakes in the first place, but they could provide victims with more robust legal means of getting them taken down, and drag our laws into the 21st century in the process. Artificial Intelligence has shown troubling signs of bias, the work of Safiya Umoja Noble’s Algorithms of Oppression (2018), showed how seemingly ‘impartial information sorting tools’ actually perpetuate systematic racism.

The tech CEO said preventing such harmful outcomes will be the responsibility of both the companies building the AI, as well as the people using the tools. The meme appears to indicate that the internet porn – as well as homemade and Cable TV porn – is dead and that generative AI will soon become the leader in the adult industry. As always, Musk, 51, was chiming in on his social media site and commented on a meme that showed the grim reaper appearing to dispose of various forms of communication, or porn mediums. Along with benefits to the industry, there are a number of ways artificial intelligence can be used in adult entertainment to mispresent or take advantage of the likenesses of individuals who did not consent to the representation. The experience of Butterworth and other Replika users shows how powerfully AI technology can draw people in, and the emotional havoc that code changes can wreak.

Online Safety Bill to clamp down on revenge porn

Brands will have to push their marketing teams even further to construct campaigns that aren’t mass-producing content by code, but use ideas that connect with people on a human level. Shallowfakes are videos manipulated using basic editing tools, such as using speed effects to show something fake. In effect, some shallowfake videos make their subjects seem impaired when they are slowed down or overly aggressive if sped up. An example of a popular shallowfake is the Nancy Pelosi video that was slowed down to make her look drunk.

This case reminds people of ZAO, a deepfake app that sparked major privacy concerns in China , the Ministry of Industry and Information Technology (MIIT) ordered its removal from app store in 2019. The long awaited Online Safety Bill, already criticised for not doing enough to crack down on image-based sexual abuse, faces delays in parliament. But another legal avenue for victims that is so far unexplored is a civil claim for breaches of data protection regulations, misuse of private information and breach of confidence against the sites that distribute revenge porn. Generative AI could also impact digital well-being through the creation of highly personalised digital experiences. We have already seen how this plays out with ad targeting and the attention-based economy of social media platforms. Very simply, GANs are two networked algorithms playing a cat-and-mouse game against each other (hence, the “adversarial”).

Eleanor Leedham argues for victims of revenge porn to receive compensation

“Much more work needs to be done to understand what harmful associations models might be learning, because if we work with human data, we are going to learn biases,” says Ghassemi. The prompt ban was first spotted by Julia Rockwell, a clinical data analyst at Datafy Clinical, and her friend Madeline Keenen, a cell biologist at the University of North Carolina at Chapel Hill. Rockwell used Midjourney to try to generate a fun image of the placenta for Keenen, who studies them. To her surprise, Rockwell found that using “placenta” as a prompt was banned. She then started experimenting with other words related to the human reproductive system, and found the same.

Founder of the DevEducation project

And even if the real video (i.e., before shallowfake alteration) is easy to locate on the Internet, the less discerning could still fall for and spread fake content without thinking twice. Such a digital watermark automatically applied onto the images would not necessarily remove some of the consequences to the victim of such an image being posted, but it would at least verify that the image is not real. With technology ever improving, there will soon be a point where it is impossible to distinguish a genuine image from one generated by AI and accordingly we will not be able to rely on a viewer identifying what is real and what is not. For example, online nudification software which virtually strips women of their clothing creates credible manipulated images. Use of nudification software is increasing at rapid pace – in 2020, a website called DeepSukebe was launched and received 38 million hits in 2021.

Social Media

Private sexual images published without consent should not be on any public websites. The fact they can be found on multiple adult content sites indicates that those sites have inadequate checks and processes for preventing the upload of illegal images, removing those that have made it onto the site and ensuring no further dissemination. Replika’s former head of AI said sexting and roleplay were part of the business model. Artem Rodichev, who worked at Replika for seven years and now runs another chatbot company, Ex-human, genrative ai told Reuters that Replika leaned into that type of content once it realized it could be used to bolster subscriptions. In July, researchers at the University of Washington developed a new machine learning tool that turned audio clips into realistic, lip-synced videos of former US president Barack Obama. Making AI use clear would go a considerable way to improving transparency, however it would not necessarily eliminate the harm of deepfake pornography that continues to appear realistic and remain online.

The gaming platform Twitch also recently updated its policies around explicit deepfake images after a popular streamer named Atrioc was discovered to have a deepfake porn website open on his browser during a livestream in late January. OpenAI says it removed explicit content from data used to train the image generating tool DALL-E, which limits the ability of users to create those types of images. The company also filters requests and says it blocks users from creating AI images of celebrities and prominent politicians.

A series that became popular in the UK included images of politicians doing low-paid, gig economy jobs with Sunak as a Deliveroo driver, Matt Hancock pushing supermarket trolleys and Liz Truss pulling pints. For all the promise of machine-learning to deliver some brilliant future, what it really does is trap us in the recent past — since it can only make its predictions from things we’ve already done. This is why Facebook is always trying to sell you the raincoat you bought last week.

Dr Emilia Molimpakis, Neuroscientist and the CEO & Co-Founder of thymia explains how the very problem of bias in AI, may help us create real-world solutions. If we decide to reward our attention and money to cheaply produced, machine-made works with no original thought behind them, then human creativity will suffer. If readers conscientiously decide to reward writers who produce quality pieces of originality, then fiction will continue to thrive and help us to explore the human condition.