EU: Microsoft-Mistral Deal Needs Further Investigation

European regulators are taking a closer look at Microsoft’s new partnership with Mistral.

The tech giant announced Monday (Feb. 26) that it planned to make the commercial version of Mistral’s artificial intelligence model available via Microsoft’s Azure cloud platform.

One day later, the European Commission said it would examine the Microsoft-Mistral deal as part of its ongoing investigation into partnerships between Big Tech and AI companies, Reuters reported.

The EU warned earlier that Microsoft’s ongoing support of OpenAI could fall under the umbrella of Europe’s merger rules, per the report.

“What is emerging shows even more that it was good not to water down our ambition on the safety of [general purpose AI] models with systemic risks, following legitimate but strong lobbying from companies like Mistral,” said Brando Benifei, a member of the European Parliament who oversaw the creation of the EU’s AI Act, according to the report.

Microsoft last year integrated Mistral 7B into its Azure AI model catalog. With its extended partnership, the company said in its Monday announcement that it is adding Mistral Large, the company’s “flagship” commercial model.

“Mistral Large is a general-purpose language model that can deliver on any text-based use case thanks to state-of-the-art reasoning and knowledge capabilities,” the company said in the announcement. “It is proficient in code and mathematics, able to process dozens of documents in a single call, and handles French, German, Spanish and Italian (in addition to English).”

Meanwhile, PYMNTS examined Monday the way the AI race “is increasingly between governments and their regulators, as nations around the world seek to become the global leader in both the safe development and responsible deployment” of the technology.

Different governments have different approaches. For example, that report said, Great Britain “is taking a decidedly ‘pro-innovation’ approach to the question of AI regulation.”

The EU’s AI Act, on the other hand, is more risk-based. Once put into effect, it will apply to every AI company in the bloc, along with users of AI systems located within the EU.

“In contrast to the EU’s risk-based categorization of AI systems, the U.K. government’s regulatory designs take an alternative approach by distinguishing between capability-based categorizations of AI systems along with outcome-based categorizations of AI risks,” PYMNTS wrote Monday.

For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.