Does the EU AI Act Require Labeling AI-Generated Cold Emails?
Updated on 26 August 2026
Short answer: for most SDR teams, no — an AI-drafted cold email that a human reviews and sends does not need an "AI-generated" label under the EU AI Act. But the details matter, some related duties are real, and this article is practical guidance, not legal advice.
What Article 50 actually requires (in force since 2 August 2026)
- Chatbots/direct AI interaction: if a person is interacting directly with an AI system (a bot conversing on its own), that must be disclosed. A templated or AI-drafted email sent by a human does not fit this category.
- Synthetic media / deepfakes: AI-generated images, audio and video require marking. Plain-text sales email is not in scope the way media is.
- The human-review carve-out: AI-assisted text that undergoes human editorial review and where a human takes responsibility is broadly exempt from generated-content labeling duties.
- The provider-side deadline (2 December 2026): providers of generative AI systems must make outputs machine-detectable (watermarking/provenance). This obligation sits on the AI providers — OpenAI, Anthropic, etc. — not on you as a user, but it means AI-drafted text may become detectable by receivers regardless of labels.
What outbound teams should actually do
- Keep a human in the loop on sequences — it's both the compliance carve-out and good deliverability practice (engagement-based filters punish obviously templated volume).
- Don't run autonomous AI reply-bots on EU threads without disclosure — a bot that negotiates or converses autonomously IS in Article 50 territory.
- Expect detectability: after 2 December 2026, assume receivers can increasingly detect machine-generated text. Quality and relevance, not disguise, is the durable strategy.
- Remember the older laws still bite harder: GDPR lawful-basis and ePrivacy rules on B2B outreach remain the bigger EU compliance risk for cold email than the AI Act.
Deliverability is the enforcement layer you feel first: authentication failures and blacklist hits kill EU outreach long before a regulator does. Grade your sending domains.
Signals disappear. Monitoring shouldn't.
Aurelius continuously verifies SPF, DKIM, DMARC, MX and 10 blacklists across your sending domains, with alerts on any regression. Free for 2 domains.
Start monitoring free
EU AI Act series: Article 50 disclosure deep-dive · AI SDR compliance checklist · More: Blacklist removal · Microsoft crackdown · Domain Grader