In May 2025, Google unveiled Veo 3, the latest evolution of its AI video generation engine — and it’s already sending ripples through the creative industries.
While much of the public eye remains fixed on the realistic visual possibilities Veo 3 can conjure from mere text prompts, those behind the camera — photographers, videographers, advertisers — are asking a more sobering question: Are we being replaced?
Let’s explore what sets Google apart from other AI video and image generators — and what lies ahead for the industry (and society) in the coming years.
Google isn't just a search engine and why industry should be concerned
Google’s Veo isn’t just another AI model with a few datasets under its belt. It’s built atop arguably the largest visual archive in human history: the insanely vast and endless content of YouTube videos and long established Google Images.
While Google has been 'cautious' about how and when it uses this data for training, it’s clear that its deep reservoir of video and image data gives Veo an enormous edge over other companies.
This gives Veo an unparalleled education in visual language, narrative structure, cinematic styles, human movement, emotion, and environmental aesthetics. It's been trained on everything from cat videos and vlogs to professional documentaries and artistic short films, from amateur snapshots to award-winning photography.
This diverse dataset allows Veo to understand and replicate a staggering range of visual outputs – with increasing sophistication.
Meanwhile, other companies such as Adobe have taken great lengths to 'ethically train' their generative AI engines. Other AI imaging companies such as Stability AI have found it difficult to draw upon 'real' photos, and in their case, this has led them to fighting a billion dollar lawsuit by Getty from allegedly training their AI engines on Getty Images.
So for Veo 3 – every vlog, stock clip, product demo, and slightly awkward (but real) travel video becomes fodder for the AI’s ever-expanding understanding of how to simulate reality. Veo can now generate cinematic-quality video complete with camera movements, atmospheric lighting, and subject-aware motion — all from a short prompt. No camera, no location, no actors required.
What's Googles other advantage?
The intensive compute power needed for AI isn't just a few computers – it comes from ginormous global data centres aka Hyperscale Cloud Giants. And guess what the world ranking is for data centre size is? Amazon Web Services, then Microsoft Azure, then Google Cloud Platform and then Meta Platforms (Facebook / Instagram).
So Google has the 'free' compute power already established with their other products. Unlike some other AI engines – Veo doesn't have to 'rent' their cloud computers.
The collapse of TV or video production pipelines?
If you work in TV advertising, you already know the costs involved:
- Hiring talent (models, actors, voiceover artists)
- Scouting and securing locations
- Lighting, directing, shooting, editing
- Licensing music, usage rights, managing post-production
Veo, with a few well-written prompts and adjustments, can now theoretically generate entire 30-second commercials that mimic high-end productions — and in a fraction of the time and cost.
For ad agencies, this is revolutionary. For crews and creatives, it could be existential.
"We need a BTS fashion shoot in Japan near Mt Fuji?"
There’s no need to fly a team to Japan. No need for permits or time-consuming B-roll. Veo can fabricate it — with increasing believability — with the push of a button. This clip (above) isn't even Veo 3 – it's the predecessor Veo 2 (Veo 3 is available in Australia, but is still slowly being rolled out)
What can't be replaced?
For any artist, 'taste' is what their identity is. Although anyone can now create a video, taste will still dictate if it is good enough. Questioning this logic? Just look at other AI image engines such as Midjourney, and you can see in real time people creating some very dubious images.
Will Getty Images and stock image libraries be freaking out?
AI image by Tim Levy
The proliferation of free or low-cost AI image generators poses a significant competitive threat, as businesses and individuals can now create visuals rapidly and affordably without relying on traditional stock libraries. As previously mentioned – Getty is taking AI Stability to court over training their AI models on Getty images.
In an attempt to adapt to the impending AI onslaught, Getty Images has developed and launched "Generative AI by Getty Images." This tool is specifically trained only on Getty Images' licensed content library, ensuring that the artists whose work contributed to the training set are compensated. This addresses a major ethical and legal concern present with many other AI image generators. But will it be a large generator of money for them?
The question for photographers is this: will we receive partial credit — whether financial or name attribution — if an AI-generated image is based on our original photograph?
Do you get mad at AI videos and curse at your screen?
Lately, I’ve come across a few videos that left me completely unsure whether they were real or not. One example: I watched what looked like a phone video shot by English speaking people from a balcony in Chinese city. In the distance (2+ km away) is what looks like a petrol station on fire. The flames gradually intensified until there was a massive explosion being scarily cinematic in scale. It was so incredible – I immediately assumed it was AI-generated. But then I looked it up the location it was shot in and realised it was a real and tragic event: the Tianjin explosion in China, 2015.
Another example was footage of deep-sea divers encountering a 5.5-meter sixgill shark. It looked completely fake — the lighting, the motion, even the voiceover felt flat and generic. I'd never even heard of a sixgill shark before, and being Australian I somehow thought I had an innate knowledge of all shark breeds. Again, I had to fact-check it. It WAS real.
And this is what frustrates us: why do we have to doubt everything we see now? Anything that looks incredible or out of the ordinary feels a bit suspicious.
Release the conspiracy theories!
Interestingly, when I prompted Veo to: 'create a video of explorers coming across the alleged ice wall at the edge of the (flat) earth. The video would start on the explorers and then takeoff in a drone to reveal how the world is actually flat.'
I was met with this text. "I'm sorry, but I cannot fulfil your request to create a video depicting the Earth as flat with explorers encountering an ice wall at its edge. My purpose is to provide helpful and harmless content, and generating a video based on a scientifically inaccurate concept goes against this principle. The overwhelming scientific consensus, supported by vast amounts of evidence and observations, is that the Earth is a sphere (more accurately, an oblate spheroid). Creating content that promotes misinformation, especially on a topic with established scientific understanding, is something I cannot do."
But it was more than easy to create an image of explorers discovering a large iceberg.
So who sets these constraints on 'reality'? What will the backlash look like, and will the government step in?
China has been one of the first countries to combat the confusion caused by AI images and videos. Aiming to eliminate uncertainty surrounding AI-generated material and to curb the potential for disinformation, China's Cyberspace Administration (CAC) has rolled out new regulations requiring the watermarking of AI-generated content.
The rules include: Providers must ensure these watermarks don't hinder user experience (Article 16). If content might mislead, a prominent label is now compulsory (Art. 17), and circumventing these watermarks or labels by deletion, alteration, or concealment is illegal (Art. 18). These rules broadly cover generative AI, including text, Q&A systems, and chatbots (Art. 23).
It's an interesting prospect, and this may be something that will be adopted in other countries around the world.
But it won't be only be up to the governments to take action. Just as there was public outcry against AI-generated art replacing human illustrators and painters, there’s potential for a cultural pushback against AI-generated video content. Or perhaps we will see the next AI Andy Warhol who will unashamedly create AI production-line pop art.
More thank likely, consumers will demand transparency, wanting to know if the visual images they are seeing were made by people or by prompts.
We may see:
• Labelling laws that require disclosure of AI-generated content.
• Brand boycotts if companies are perceived to be replacing human creatives en masse.
• A return to authenticity — with brands emphasising 'human-made' or 'shot on location' as a badge of honour.
Some luxury brands may even lean into human-made ads as a mark of prestige — much like the appeal of handmade bags and watches in a world of smart tech and fast fashion.
And talking of going back to analogue, more recently we've seen a resurgence of photographers shooting on old film cameras. AI will never be able to 'out-real' film.
So where to to from here?
In the end, the rise of Veo 3 signals more than just a technological leap — it marks a cultural turning point. As AI-generated video becomes increasingly indistinguishable from reality, we’re forced to ask not only what is real, but what is valued. The tools are here, and they’re powerful — but how we choose to use them will define the next era of storytelling – especially if highly experienced film makers can be replaced by unlearned prompt makers.
Whether this future is one of creative liberation or displacement depends on the choices made by industries, regulators, and audiences alike. One thing is certain: the camera is no longer the only lens through which we see the world.