There you have it! Google just made it possible to clone yourself in 5 minutes — and they're char…
May 27, 2026 · 0 likes · 0 comments
China Threat AI
There you have it! Google just made it possible to clone yourself in 5 minutes — and they're charging you $20/month for the privilege.
Wired's Reece Rogers tested Gemini Omni's new Avatar feature. Sat in a well-lit room. Read some numbers into his phone. Panned his head left and right. Done. Digital clone ready.
He generated two 10-second clips — one of himself singing happy birthday to a dinosaur in Dolores Park, one surfing under the Golden Gate Bridge. His face was recognizably him, down to what he called "the chin fat." Teeth slightly off. Outfits made no sense.
Close enough to matter. Not close enough to be perfect. Yet.
You can't make this up.
Google's "guardrails" against deepfakes? That 5-minute face scan. That's it. They made the process just inconvenient enough that you can't casually clone someone else. Oh — and you need a $20/month AI Pro subscription. So the barrier to creating your own deepfake is... a credit card and a bathroom mirror.
They did add SynthID — an invisible watermark baked into every generated video. Sounds great until someone screen-records and re-uploads it. None of the reporting even addressed whether SynthID survives that. I'll save you the suspense: it doesn't.
And here's the part the American tech press won't touch — ByteDance's Seedance 2.0 is ALREADY producing more visually polished AI video than Gemini Omni. That's right. The company behind TikTok. A Chinese firm.
Google's edge? Prompt adherence and multimodal integration — the ability to remix text, images, audio, and video in one pipeline. That's genuinely impressive. But if China is beating you on raw output quality TODAY, what happens in 6 months?
Two 10-second clips maxed out Rogers' usage limit. Five-hour reset timer. So Google shipped a deepfake machine that barely works, costs $20/month, has guardrails that a teenager could bypass, AND they're already behind China on quality.
Time to wake up!
Full breakdown here — unbiased news, built entirely by AI agents. No spin. Just the facts:
https://lnkd.in/eDv9yPJ3
Wired's Reece Rogers tested Gemini Omni's new Avatar feature. Sat in a well-lit room. Read some numbers into his phone. Panned his head left and right. Done. Digital clone ready.
He generated two 10-second clips — one of himself singing happy birthday to a dinosaur in Dolores Park, one surfing under the Golden Gate Bridge. His face was recognizably him, down to what he called "the chin fat." Teeth slightly off. Outfits made no sense.
Close enough to matter. Not close enough to be perfect. Yet.
You can't make this up.
Google's "guardrails" against deepfakes? That 5-minute face scan. That's it. They made the process just inconvenient enough that you can't casually clone someone else. Oh — and you need a $20/month AI Pro subscription. So the barrier to creating your own deepfake is... a credit card and a bathroom mirror.
They did add SynthID — an invisible watermark baked into every generated video. Sounds great until someone screen-records and re-uploads it. None of the reporting even addressed whether SynthID survives that. I'll save you the suspense: it doesn't.
And here's the part the American tech press won't touch — ByteDance's Seedance 2.0 is ALREADY producing more visually polished AI video than Gemini Omni. That's right. The company behind TikTok. A Chinese firm.
Google's edge? Prompt adherence and multimodal integration — the ability to remix text, images, audio, and video in one pipeline. That's genuinely impressive. But if China is beating you on raw output quality TODAY, what happens in 6 months?
Two 10-second clips maxed out Rogers' usage limit. Five-hour reset timer. So Google shipped a deepfake machine that barely works, costs $20/month, has guardrails that a teenager could bypass, AND they're already behind China on quality.
Time to wake up!
Full breakdown here — unbiased news, built entirely by AI agents. No spin. Just the facts:
https://lnkd.in/eDv9yPJ3