If you deploy it, you own it
Spain issues a checklist for AI agents, Canada wants buyers of AI held accountable too and four UK regulators say today's rules already apply. Whoever deploys AI stays responsible for it.
This week is about responsibility staying where it belongs. Regulators in Spain, Canada and the UK each make the same point in their own way: if you deploy or buy AI, you remain accountable. Anthropic’s IPO also shows that control at a major AI vendor can work differently from what you might assume.
-
Agencia Española de Protección de Datos (AEPD) ·
La AEPD publica un decálogo para delegar tareas en asistentes y agentes de IA sin perder el control sobre los datos personales
Spain's data protection authority wrote these ten points for individual users, but they work just as well as a checklist for agents inside an organisation: only the access a task needs, no "always allow" and a confirmation before an agent sends or pays. Ask the same of your vendors, including a record of what the agent did and a way to stop it at once.
-
TechCrunch ·
Who's really in charge at Anthropic after the IPO?
After its IPO, ordinary Anthropic shareholders will have little say. An independent trust elects a majority of the board and the founders jointly hold 50.1% of the voting power. If you build on this vendor, you should know who sets its direction. Treat that as part of vendor risk, next to price and technology.
-
Office of the Privacy Commissioner of Canada ·
Canada's federal, provincial, and territorial privacy commissioners and ombuds call for the responsible use of artificial intelligence in health and health research
Canada's privacy regulators want a legal duty for everyone who develops, procures or uses AI in health care to have a governance and accountability framework. Note the word procures. Buying AI does not hand the responsibility to the vendor, and that holds well beyond health care.
-
Digital Regulation Cooperation Forum (DRCF) ·
Protecting people and businesses in an agentic economy: A Joint Statement by DRCF Members
The CMA, FCA, ICO and Ofcom state together that existing legal and regulatory frameworks already apply to personal AI agents. There is no separate regime to wait for. If your agents act on behalf of customers, test them now against the rules you already know.