California’s AI Transparency Act went into effect on January 1, 2026.
If your platform has more than 1 million users and you’re using generative AI, you’re required to watermark AI-generated content.
Are you doing it?
Most companies aren’t.
And in 7 months, the EU AI Act’s high-risk enforcement provisions kick in.
August 2, 2026.
If you thought you had time to figure this out, you were wrong.
What Just Happened (While You Weren’t Paying Attention)
California SB 942 became law on January 1st.
Here’s what it requires:
If you operate a generative AI system with more than 1 million monthly users in California, you must watermark AI-generated content.
If you provide tools that let users create or modify content, you must disclose when AI is involved.
Who this affects:
SaaS platforms with AI content generation. Social media platforms using AI. Content platforms with AI tools. Marketing platforms generating AI content.
If your users are creating AI-generated content on your platform and you’re not watermarking it, you’re violating California law right now.
Why Most Companies Ignored This
I’ve talked to dozens of General Counsels in the past month.
Almost none of them implemented SB 942 compliance before January 1.
“We didn’t think it would actually pass.” It passed in September 2024. You had 15 months.
“We thought implementation would be delayed.” It wasn’t.
“We’re still evaluating technical solutions.” You should have been evaluating 6 months ago.
“We’re waiting to see how it’s enforced.” That’s not a compliance strategy. That’s hoping you don’t get caught.
Most companies with generative AI platforms entered 2026 already non-compliant.
The August Cliff: EU AI Act High-Risk Enforcement
August 2, 2026 is when the EU AI Act’s high-risk requirements become enforceable.
That’s 7 months from now.
What qualifies as high-risk AI:
AI used in employment decisions such as hiring, promotion, termination.
AI used in education such as student evaluation, admissions.
AI for credit decisions or credit scoring.
AI for critical infrastructure management.
AI that profiles individuals for risk assessment.
If your AI falls into any of these categories and you have EU users, you’re subject to the EU AI Act.
What compliance requires by August 2:
Risk management systems throughout the AI lifecycle.
Technical documentation of the AI system.
Record-keeping and logging capabilities.
Human oversight mechanisms.
Accuracy, robustness, and cybersecurity measures.
Conformity assessments before deployment.
That’s not a policy document. That’s operational infrastructure that takes months to build.
Who’s Most Exposed Right Now
SaaS platforms with generative AI features. You added AI to stay competitive. You probably didn’t think about watermarking or EU risk classification.
AI-first companies that moved fast. You built and shipped quickly. Compliance was “something we’d handle later.” Later is now.
HR tech platforms with AI. If you use AI for hiring decisions, you’re definitely high-risk under EU AI Act. August 2 is your deadline.
Enterprise platforms serving EU customers. If your AI processes EU citizen data or makes decisions affecting EU individuals, you’re in scope.
What You Should Be Doing Right Now
This Week: Complete your AI system inventory. Which systems have 1M+ users? Which are high-risk under EU rules? Assign executive ownership with authority and budget.
This Month: Select technical solutions for watermarking. Identify gaps in EU AI Act compliance. Build project plans with completion dates well before August 2.
Next 3 Months: Implement watermarking for California. Build risk management systems for EU high-risk AI. Create documentation and logging infrastructure.
Months 4-6: Complete conformity assessments. Fix issues identified in testing. Update customer communications and terms.
Month 7: Be compliant. Not scrambling. Compliant.
Here’s What You Need to Ask Yourself
Can you realistically be compliant with EU AI Act high-risk requirements by August 2, 2026?
If the answer is yes and you have a credible plan with resources allocated, execute.
If the answer is “maybe” or “we’re not sure,” you need external help immediately.
If the answer is no, you have three options: deprioritize high-risk AI features until you can comply, accept non-compliance and prepare for enforcement, or find a way to get to yes by bringing in expertise and resources.
None of these options are great. But they’re better than pretending you have time you don’t have.
Stop Waiting for Permission to Panic
California compliance is already overdue. EU compliance is 7 months away.
If you’re not in active implementation right now, you’re not going to make it.
Schedule a 30-minute AI Compliance Reality Check where we’ll:
- Review your AI systems against California and EU requirements
- Show you exactly what’s required for compliance
- Give you an honest assessment of whether you can meet the August deadline
- Provide a roadmap for what needs to happen between now and then
We’ve helped dozens of companies navigate California SB 942 and EU AI Act compliance.
Some started early and are in good shape. Others are scrambling.
The difference isn’t capability. It’s when they started taking it seriously.
The time to start was 6 months ago.
The second-best time is right now.
Brian Alexander
Founding Partner, VisionAI+ Consulting Group
Former General Counsel, AI Engineering Firm
Certified ISO 42001 Lead Auditor