Anthropic's implementation of watermarking illustrates the "Brussels Effect" in AI. To comply with the EU's AI Act, companies are building regulatory features into their core models. This results in de facto global regulation, as it's often easier than creating region-specific versions of their technology.

Universal safety filters for "bad content" are insufficient. True AI safety requires defining permissible and non-permissible behaviors specific to the application's unique context, such as a banking use case versus a customer service setting. This moves beyond generic harm categories to business-specific rules.

Company-wide learning platforms are insufficient for pharma because they teach general AI principles. They fail to address the specific, high-stakes judgments required in functional workflows, like deciding if a generated regulatory summary is defensible or a promotional claim is substantiated, leaving employees with unanswered questions for their roles.

Training that raises employee enthusiasm for AI tools without embedding deep, contextual judgment is counterproductive. It increases AI use in high-stakes workflows faster than the organization's ability to control for errors, creating a risk funded by the training budget itself and leaving the company more exposed than before.

The EU AI Act's scope is extraterritorial and applies to any company, including US-based ones, if the output of their AI system is used within the EU. This jurisdiction follows the output, not the company's location, making its reach even broader than GDPR.

Unlike US firms performing massive web scrapes, European AI projects are constrained by the AI Act and authorship rights. This forces them to prioritize curated, "organic" datasets from sources like libraries and publishers. This difficult curation process becomes a competitive advantage, leading to higher-quality linguistic models.

AI's biggest risk is not incompetence, but its tendency to fill context gaps with general industry knowledge. This can seem insightful but leads to hallucinations. AI Stewards must provide specific business data and knowledge to constrain the AI and ensure relevant, accurate output.

Simply providing data to an AI isn't enough; enterprises need 'trusted context.' This means data enriched with governance, lineage, consent management, and business rule enforcement. This ensures AI actions are not just relevant but also compliant, secure, and aligned with business policies.

Mandating training modules to boost AI competency is ineffective. It encourages passive behavior, similar to HR compliance training. True competency is only built and measured through hands-on experience and applying the tools to solve real business problems, not through completion certificates.