In an interview at the show with Rich Belcher of First Sight Media, El Gazzar stressed that event planners face a tougher order than ever because everyone in the audience (especially younger ones) wants to feel like their needs are being met, demanding a more personalized, unique and interactive experience. 

With a nod to the growing importance of AI for the events biz, Recordly.ai impressed the Event Tech Live judges and industry professionals with its cutting-edge audio transcription technology, earning the coveted Gold award. 

Remember the computer virus battles? In their heyday (1990-2000) anti-virus protection software from Symantec, Kaspersky, Avast and TrendMicro was raking in fortunes playing whack-a-mole with computer viruses that were being unleashed on PCs. Yes, we still have viruses, but thanks to myriad technologies, they are more easily detected and isolated, and much of the antivirus protection is built into our devices and on the cloud, making virus protection less onerous. Fast forward two decades. Deep fakes are the new “virus”. 

Deepfakes, digital creations of people, places and things that are indistinguishable from the real thing, are this decade's “virus” challenge. AI tools are both exacerbating and trying to solve the problem at the same time.

How many have you seen? Taylor Swift hawking Le Creuset cooking ware, Tom Hanks endorsing a dental plan, Joe Biden’s convincing robocall. A high school principal’s racist remarks. Kids are deepfaking each other, their teachers and parents, and often for hurtful reasons. And deepfakes come in all media forms: voice, photos and videos. 

For a hot minute, deepfakes were easy (and sort of fun) to try and spot just by looking at them. In videos, the fakes didn’t have breathing chests, the pupils of their eyes looked vacant, their eyes didn’t blink normally, their lip syncs were off, reflections and aspect ratios were off, hands were malformed. Thanks to more training, AI now has improved its ability to simulate likenesses. The difference between fake and real is getting damn hard to judge. We need to become art forgery experts! 

What’s being done to combat deepfakes? AI’s are being trained on fakes to make them easier to identify. They look for things like subtle patterns of irregular pixels, tone modulation in voice, and blood flow through the skin for example. FakeCatcher, developed by Intel, looks for subtle biometric data in the pixels of a video that makes us human. Sentinel, another deepfake identifier, is trained to examine facial expressions, blinking patterns, audio manipulation, and natural language processing. New deepfake detectors are being announced daily. 

Digital watermarking, another software technique, embeds information about the source of an image, video or audio, and lets software track the source to make sure it’s being used and attributed properly. Unlike the watermarks you see stamped on images at sites like Getty and Shutterstock, these are invisible digital watermarks that usually come in the form of embedded code and typically use stenographic techniques and extraction algorithms to track the asset. 

Not saying that all digital identities are bad things, either. There’s a lot to be said about owning your digital identity and using it as a source of royalties and income. Say Tom Hanks really wants to do a film, a video, an animated voice film and a TV commercial at the same time. Hand (Human and Digital) has created a system for registering and verifying digital likenesses, and tracking their monetization. Only big talent agencies and celebs will afford it at first, but it’ll trickle down to us mere mortals eventually.

But software isn’t the only solution. Regulations and technical standards play their part. Governments are enacting regulations surrounding deepfakes. (HInt: The US is lagging behind the UK, EU, China and Singapore.) The Coalition for Content Provenance and Authenticity (C2PA) is an industry group of heavy hitters, including Adobe and Microsoft, that’s addressing the prevalence of misleading information online through the development of technical standards for certifying the source and history (or provenance) of media content. 

TikTok, YouTube and Meta are all beginning to require some sort of labeling of AI content. Fearing election deepfakes, OpenAI launched a new deepfake detector for disinformation researchers, designed to help identify fake content generated by its own DALL·E 2 image generator.

Once again, technology forces us to play cops and robbers, Like computer viruses before it, it’s unlikely there will be one silver bullet to solve the deepfake problem but a suite of solutions working together may help control the damage.

Real or Fake? Northwestern Kellogg School has a test you can take to see how well you can differentiate and to what degree of certainty. Then they’ll tell you how well you compare to others. 

I’ll be speaking about deepfakes with an illustrious panel of guests at the AI for Good Summit sponsored by ITU in Geneva at the end of the month.

SXSW on the Thames 
It’s doubtful that the BBQ will be as good, but it’s great news that Austin’s signature event is heading to London in June of 2025. 

Open Sauce: The Shape of Events to Come?
No celebs, lots of makers and creators, lots of interactive activations, and plenty of time for networking. Can new events like Open Sauce succeed?