Google's SynthID works, but labels will not solve trust
Ars Technica AI
30 July 2026
How this applies to you
If your business publishes images, ads, training material, or customer-facing guides, assume people will ask what is real. Keep source files, approval notes, and a simple disclosure policy rather than relying only on detection tools.
Ars Technica tested Google's SynthID watermark and found it hard to break. The report also makes clear that labelling AI content will not, by itself, fix online trust.
Why it matters. If your business publishes images, ads, training material, or customer-facing guides, assume people will ask what is real. Keep source files, approval notes, and a simple disclosure policy rather than relying only on detection tools.
