Revisión de información Spotify
26 September 2026 · Issue #15
Hola,
Queremos asegurarnos de que tu información esté actualizada.
Por favor, revisa tus datos cuando tengas un momento. El proceso es rápido y sencillo.
Puedes acceder a la página de revisión mediante el siguiente enlace.
Gracias.
The law your chatbot is breaking
On 2 August 2026, Article 50 of the EU AI Act came into force. The requirement is not complicated: if a customer is talking to a chatbot or AI agent, they must be told so, clearly, at the start of the interaction. Not buried in a terms page. Not implied by a robotic tone. In the chat surface itself, before anything else happens.
The fines are up to €15 million or 3% of global annual turnover, whichever is higher. That's not a nuisance penalty. For a mid-size contact centre operation serving European customers, it's a material business risk. And the Act reaches beyond EU-headquartered companies — if your AI system produces outputs used in the EU, you're in scope regardless of where you're incorporated.
The practical gap is striking. According to a 2026 Gartner survey, 88% of contact centres now use some form of AI. Most of those deployments went live before anyone had an EU compliance checklist. Regulators have already signalled that a vague label like "assistant" or a disclosure buried in metadata won't satisfy the obligation. The disclosure has to be perceivable in the interaction itself.
There's one wrinkle worth knowing. Systems already on the market before 2 August have until 2 December 2026 to meet the machine-readable marking requirement under Article 50(2). But the basic chatbot disclosure obligation — telling users they're talking to AI — applies immediately, to everyone, with no grace period. That distinction matters and most compliance summaries are blurring it.
$46,000
That's the true per-agent cost of turnover in a contact centre, according to Insignia Resources, once you factor in recruiting, training, and the 6-to-8-month ramp before a new hire reaches the productivity of the person they replaced. Direct replacement costs run $10,000 to $20,000. The rest is lost output. Industry-average annual attrition is 40-45%. For a 100-seat operation, that's potentially $1.8 million to $4.6 million walking out the door each year — a figure that dwarfs most AI implementation budgets and gets almost no boardroom attention.
88% adopted it. 25% use it.
CMSWire's 2026 contact centre research lands on a number that should embarrass every vendor in this space: 88% of contact centres report using AI, but only 25% have fully integrated it into daily workflows. Three in four organisations bought the technology and didn't operationalise it.
The industry keeps blaming the technology. The actual problem, per Gartner's 2025 AI Implementation Survey, is that 62% of underperforming AI support projects fail because of bad data preparation, not the model. You can't automate a knowledge base that was already broken. The AI just breaks it faster and at scale.
Meanwhile, 37% of organisations using AI agent assist tools report measurable reductions in agent turnover. That's the use case people keep burying in footnotes. Fixing the attrition problem with AI saves more money than the deflection rate slide in your vendor's deck. Nobody's leading with that number.
Re-reading the original EU AI Act text this week. It's clearer than people give it credit for. The confusion isn't in the law — it's in the compliance industry that grew up around it.