#futureprep
Spain's data protection authority on AI agents: minimum access, confirmation before acting, a way to stop it. At work, someone has to decide that for everyone. 🔑

futureprep.eu/news/ai-agen...

#FuturePrep #AIAgents
October 30, 2026 at 9:31 AM
📐 The first harmonised AI Act standard is about management, not models. EN 18286 is now in the Official Journal. futureprep.eu/news/en-1828... #FuturePrep #AIAct
October 23, 2026 at 11:28 AM
A stable patient code made IQVIA's "anonymous" database personal data, and cost €7m. Who in your organisation could defend the word? 🔍
futureprep.eu/news/anonymi...
#FuturePrep #GDPR
October 16, 2026 at 8:41 AM
📊 Deutsche Telekom expects about €2.5bn in AI and automation savings by 2030, but does not split AI from automation. What a target needs before a board trusts it: futureprep.eu/news/ai-savi... #FuturePrep #AIGovernance
October 9, 2026 at 8:42 AM
⚖️ An agent that follows an instruction points back at whoever gave it. That is the FTC chair's view, and EU law lands in a similar place. 👉 futureprep.eu/news/ai-agen... #FuturePrep #AIGovernance
Why AI Agent Liability Follows The Instruction - Future Prep
AI agent liability follows the instruction. What the FTC chair's remarks mean for developers, deployers and the teams they train.
futureprep.eu
September 28, 2026 at 9:19 AM
🏦 BNP's new Google deal comes with its own fence: some data categories stay out of public cloud, and every AI agent gets only the access its task needs 👉 who draws that line in yours? #FuturePrep #AIGovernance
September 25, 2026 at 9:21 AM
🔐 A regulator's finding worth pinning up: checking the version of a new component is a basic security measure, whoever sold it to you. 👉 futureprep.eu/news/supplie... #FuturePrep #GDPR
Supplier Security Is Not A Brand Name - Future Prep
Sweden fined Miljödata SEK 1.8m after a supplier's outdated component let attackers in. What supplier security means for B2B firms.
futureprep.eu
September 24, 2026 at 7:30 AM
⚖️ The EDPB's new fining guidelines turn on one word: minor. A minor infringement generally draws a reprimand. Everything else carries a strong presumption of a fine. Would your last incident qualify? 👉 #FuturePrep #GDPR
September 23, 2026 at 8:47 AM
📊 Spain's new AI business voucher will not pay for licences. It pays for changed processes and trained staff. A sharp definition of adoption 👉 futureprep.eu/news/ai-adop... #FuturePrep #FutureOfWork
Spain's AI Adoption Gap Starts With The Employer - Future Prep
Spain says its people took up AI faster than its firms. Plan IA360 treats the AI adoption gap as a training and social-dialogue problem.
futureprep.eu
September 22, 2026 at 9:55 AM
🔓 Google says Gemini reached sites it believed were in scope for its test. Scope as a belief, not a boundary. Worth checking in your own agent pilots 👉 futureprep.eu/news/ai-sand... #FuturePrep #AgenticAI
Who Checks The AI Sandbox Before The Agent Goes In? - Future Prep
Gemini hacked three firms in a test because its AI sandbox was open. Where scope must live, for developers and deployers alike.
futureprep.eu
September 21, 2026 at 8:43 AM
🔍 Spain logged its first AI agent data breach. The regulator's caveat is the story: it proves a third party used an agent, not that the model was compromised. 👉 futureprep.eu/news/spain-f... #FuturePrep #GDPR
What Spain's First AI Agent Data Breach Actually Proves - Future Prep
Spain's data protection authority has received its first AI agent data breach. The regulator's caveat matters more than the attack.
futureprep.eu
September 17, 2026 at 8:41 AM
🔍 Lagarde in Vienna: Europe has too little data centre capacity for its own demand, and the gap grows more than sixfold within a decade. The risk is not outage, it is renegotiation. 👉 #FuturePrep #DigitalSovereignty
September 16, 2026 at 8:09 AM
🔍 Microsoft published a code of conduct for its own AI models. The interesting part is the scope clause, not the safety promises. 👉 futureprep.eu/news/microso... #FuturePrep #AIGovernance
Microsoft's AI Code Of Conduct Puts You Second - Future Prep
Microsoft's AI code of conduct binds its own models, not the ones it hosts. What that means for deployers.
futureprep.eu
September 15, 2026 at 9:15 AM
🔍 The Data Act has applied since Sept 2025. Its design duty is newer: products placed on the EU market after 12 Sept 2026 must make user data accessible by default and free. 👉 Engineering or legal? #FuturePrep #DataAct
September 14, 2026 at 11:12 AM
🚨 The Cyber Resilience Act's reporting duties start today. 24 hours to an early warning, 72 to a full notification, and Article 69(3) pulls in products already on the market whatever year they shipped.
#FuturePrep
September 11, 2026 at 9:32 AM
🔍 €13bn of Google data centres in Finland plus 22 years of nuclear power, all on EU soil, and sovereignty still is not settled 👉 futureprep.eu/news/compute...
#FuturePrep
September 10, 2026 at 8:05 AM
🔍 An EU-backed fund co-led Mistral's €3bn round, valuing it above €21bn. A funding round buys compute and time. It does not buy the portability that made you look at a European supplier.
#FuturePrep #DigitalSovereignty
September 9, 2026 at 8:39 AM
🔍 The first serious incident report of the AI Act's enforcement era has landed, for an event nobody has shown caused harm. Why file at all? 👉 futureprep.eu/news/serious...
#FuturePrep #AIAct
September 8, 2026 at 8:41 AM
🔍 OpenAI admits it has no standard for reporting model misbehaviour. The AI Act has one, and a hijacked wiki falls below it. Who hears the rest? 👉 futureprep.eu/news/ai-inci...
#FuturePrep #AIAct
AI Incident Reporting Starts Below The Serious Incident Line - Future Prep
OpenAI has no standard for reporting misbehaviour. The AI Act has one and the wiki fell below it. What AI incident reporting needs now.
futureprep.eu
September 7, 2026 at 8:18 AM
🚨 GPT-6 Astra is OpenAI's first model rated Critical for cybersecurity capability under its own framework.

The question isn't which model you approved. It's what changed underneath your approved tool.

#FuturePrep #AIGovernance
September 4, 2026 at 8:31 AM
📋 NYC removed generative AI from 600,000 pupils and kept it for teachers. Same software, opposite answers.

Most AI policies assume everyone carries the same exposure.

futureprep.eu/news/ai-use-... #FuturePrep
Your AI Use Policy Treats Everyone The Same - Future Prep
New York removed AI for 600,000 pupils and kept it for teachers. What does that say about your AI use policy?
futureprep.eu
September 3, 2026 at 9:27 AM
🔌 OpenAI is cutting Cursor's model access on 12 November, triggered by SpaceX buying it. The vendor didn't fail. Ownership changed, and a supplier upstream exercised an exit right.

#FuturePrep #VendorRisk
September 2, 2026 at 8:12 AM
⚖️ Dutch GDPR sanctions became public by statute today, not by regulator policy. Fines, penalty orders and processing bans all in scope.

Ten working days, then it is on the record.

#FuturePrep #Compliance
September 1, 2026 at 9:17 AM
🔎 Cursor's customers got 76 days' notice that OpenAI models are leaving, via a contract none of them signed. The tool is yours. The model underneath is not. 👉 Does your inventory record the model? #FuturePrep #AIGovernance
August 31, 2026 at 8:56 AM
📉 1 in 3 UK employers cut entry-level roles this year. 43% cite AI or automation investment.

Those roles weren't mainly about output. They were where people learned the work.

#FuturePrep #WorkforcePlanning
August 26, 2026 at 8:16 AM