The EU's proposed GDPR amendments won't take effect until 2031. Current enforcement is not waiting. Cumulative fines have hit €5.88 billion, and the frontline priorities right now are cookie consent dark patterns and legitimate interest assessments for AI tools. If you run outbound with any AI in the stack, those two items are your 2026 compliance work.
The reform is real. The timeline is not your friend.
The proposals floating around Brussels are sensible on paper. SME exemptions, standardised cookie banners, a bit of breathing room for small operators. If you're a two-person sales team, that sounds like relief.
It isn't relief. It's a 2031 problem being legislated in 2026. Between now and then, regulators are enforcing the current rules with the current penalties. We've written before about the €5.88B fine total and the SME exemption carve-out, and the pattern hasn't changed: the exemption is a proposal, the fines are a fact.
So the practical question isn't "what will the law be in 2031." It's "what are regulators actually chasing in 2026." Two things, mostly.
Cookie consent is the cheapest fine you'll ever avoid
Dark patterns in cookie banners are the most-enforced GDPR issue in the EU right now. Not because they're the most damaging, but because they're the easiest to prove. A regulator opens your site, clicks around for ninety seconds, and either you have a compliant banner or you don't.
The patterns that get flagged:
- A pre-ticked "accept all" box, or an accept button that's visually dominant while "reject" is buried in grey 11px text.
- No way to reject without going through a second screen or a settings maze.
- "Legitimate interest" toggles pre-enabled for advertising cookies.
- A banner that reappears on every page load until you accept.
- Cookie walls that block content until consent is given.
None of these are ambiguous. All of them have been fined. If your marketing site runs on a template with a stock cookie plugin, there's a decent chance you're running at least one of them right now.
The fix is boring and takes an afternoon. Equal-prominence accept and reject buttons on the first layer. No pre-ticked boxes. A persistent settings link. A consent log that records what each visitor chose and when. If you use a CMP, check the default configuration — most ship with dark patterns enabled because they convert better.
We audited our own site last quarter and found two of the five. Neither was intentional. Both were defaults.
Legitimate interest is the assessment nobody wrote
The second enforcement priority is AI tools processing personal data under legitimate interest. This is where most B2B sales teams are exposed, and almost none of them know it.
Here's the pattern. You buy an AI prospecting tool. It enriches contact records, scores inboxes, drafts emails. Under the hood, it's processing personal data — names, emails, job titles, sometimes inferred attributes. The lawful basis your vendor relies on is usually legitimate interest.
Legitimate interest isn't a free pass. It requires a documented assessment: what the interest is, why processing is necessary, and a balancing test against the individual's rights. If a regulator asks for yours and you don't have one, "the vendor said it was fine" is not a defence.
We've covered why AI scrutiny is the real threat to cold email, and this is the mechanism. The fine isn't for sending cold email. It's for processing personal data without a documented basis.
What a legitimate interest assessment actually looks like for a small sales team:
- One page per tool. Name the tool, the data categories it touches, and the processing purpose.
- The interest: usually "B2B commercial outreach to relevant business contacts."
- The necessity test: why this processing, why this data, why not less.
- The balancing test: reasonable expectations of the data subject, safeguards in place, opt-out mechanism.
- A date and a name. Someone owns it.
That's it. It's not legal theatre. It's a document that says you thought about it. Regulators care more about the thinking than the prose.
If your AI vendor won't tell you what data they process or what basis they rely on, that's your answer. Don't buy the tool.
What this means for your outbound stack in 2026
Two practical moves, both cheap, both doable this month.
First, fix the cookie banner. Not because the reform is coming, but because the enforcement is here. It's the lowest-effort compliance win available to you.
Second, write the legitimate interest assessments for every AI tool in your stack that touches personal data. One page each. Date them. Keep them somewhere you can find in five minutes if asked.
Neither of these will make you a better salesperson. Both will keep you out of the enforcement queue while everyone else waits for 2031.
And if you're running outbound across multiple jurisdictions, the same discipline applies. We wrote about the single compliance rule that works across 20 US state privacy laws — the principle is the same. Document your basis, keep it current, don't rely on defaults.
What we'd do next
Open your site in an incognito window and try to reject cookies in under three clicks. If you can't, fix it this week. Then list every AI tool that touches a prospect's personal data and write a one-page assessment for each. If you want a platform that keeps the human in the loop on every send — and documents what it does with prospect data — give MiraReach a try.
— Mira