Two regulatory deadlines are approaching for French businesses between 2026 and 2027: e-invoicing and the European AI Act. What they have in common is less about technology than the operational burden they impose. Measuring the gap between obligations already in force and those still to come helps identify priority projects and adapt your business without having to improvise.
Regulatory timeline: e-invoicing and the AI Act side by side
Both reforms are being phased in, but their timelines do not overlap. E-invoicing distinguishes between receiving and issuing invoices: the requirement to receive electronic invoices already applies to all businesses subject to VAT, while SMEs, small businesses and micro-businesses must prepare to issue them by the 2027 deadline. Micro-entrepreneurs are not automatically exempt from the scheme because they benefit from the VAT exemption threshold.
The AI Act, meanwhile, imposes requirements on businesses that use AI, even if they do not develop it. The first obligations include ensuring staff are AI-literate and, for certain uses, being transparent with people who interact with a chatbot or view AI-generated content.
| Criterion | E-invoicing | AI Act |
|---|---|---|
| Scope | All businesses subject to VAT in France | All businesses using or deploying AI systems in the EU |
| Obligation already in force | Receiving electronic invoices | Ban on unacceptable-risk AI practices; AI literacy for staff |
| Next major deadline | Mandatory issuance for SMEs, small businesses and micro-businesses in 2027 | Phased application based on risk levels |
| Main burden | Platform interoperability and accounting system configuration | Auditing actual AI use, training and transparency |
| Underestimated risk | Rejections and duplicate invoices caused by addressing issues between platforms | Unknowingly using third-party tools classified as “high risk” |
Articles published in Blog Too’s business section regularly explore these regulatory changes and their implications for business leaders.
E-invoicing: interoperability issues the guides don’t cover
Most content about e-invoicing outlines the timeline and requirements. The reality on the ground is different. Since the rollout began in France, interoperability issues between platforms, addressing errors and configuration problems have caused rejections and duplicate invoices at some businesses.
The entire process—from supplier to e-invoicing platform to accounting software—should be tested under real-world conditions before the issuance deadline. Waiting until the last minute to discover that an invoice format is rejected or an addressing ID points to the wrong recipient can lead to cash flow bottlenecks.
Common pain points identified in practice
- Invoice formats vary depending on the platform selected, and platforms do not all handle the same standards in the same way. Testing the entire process helps identify incompatibilities before they hold up a payment.
- Addressing relies on a centralized directory, but configuration errors (the wrong SIREN number or the wrong platform listed) can mean invoices never reach the intended recipient.
- Micro-entrepreneurs benefiting from the VAT exemption threshold still need to register on a platform and prepare to issue invoices—something many are still unaware of.
For a business managing multiple suppliers, testing the entire invoicing process is a preventive investment, not an administrative formality.
AI Act: audit your use of AI before thinking about compliance
The most common mistake is to assume that the AI Act only applies to software developers. The AI Act also applies to businesses that use AI without developing it. A customer service chatbot, a CV screening tool or a supplier scoring system: each of these uses may fall into a different risk category under the regulation.
The phased rollout means businesses need to examine how they actually use AI, not just the technology they have purchased. An off-the-shelf tool may be classified as “high risk” if it is used in recruitment, credit or security.
Three practical questions to ask about every AI tool in use
Staff who use these tools must be trained. Transparency toward end users (customers and candidates) is already required for certain systems. A structured internal audit covering three areas can help map your exposure:
- What type of decision does this tool influence (recruitment, credit, access to a service, or simply a product recommendation)? The answer determines the regulatory risk level.
- Do people interacting with the system know they are communicating with AI or viewing AI-generated content? If not, the transparency requirement already applies.
- Has the tool provider documented the risk level and planned compliance measures? Without this documentation, the business using the tool bears some responsibility.
Business strategy ahead of regulatory deadlines: prioritize by operational impact
E-invoicing and the AI Act require different kinds of effort. The former is a technical and accounting project with fixed deadlines. The latter calls for ongoing audits that evolve as business practices change.
For a small business or SME, e-invoicing primarily involves the finance team and accounting provider. The AI Act, by contrast, affects marketing, HR and customer service teams, often without their knowledge.
The difference between these two reforms is also reflected in the human resources they require. E-invoicing calls for one-off configuration (followed by maintenance), while AI compliance requires ongoing monitoring of the tools in use and their regulatory classification. A business that purchases a new scoring or content-generation tool must reassess its position with every acquisition.
Sustainable business strategies are no longer based solely on marketing or product development. Regulatory compliance management is becoming a competitive advantage for businesses that prepare early, compared with those forced to make adjustments in a hurry.
