Key facts
- France's domestic intelligence agency is ending its decade-long contract with Palantir.
- France is opting for French firm ChapsVision to replace Palantir.
- France will invest €655 million in artificial intelligence.
- France will create a common chatbot for all state services.
- A public health chatbot for the Ameli agency is planned.
- A study found Mistral AI models scored below 40% in detecting Kremlin talking points.
- Some Mistral AI model answers referenced pro-Russian sources like RT and Sputnik.
- The G7 is set to discuss the future of artificial intelligence.
- Two Anthropic AI models were deactivated prior to the G7 discussions.
France is making significant moves to enhance its artificial intelligence sector and assert digital sovereignty. The nation's domestic intelligence agency is terminating its ten-year contract with U.S. AI company Palantir, choosing instead to partner with French firm ChapsVision. This decision is driven by a desire to achieve greater digital autonomy and mitigate strategic dependencies on foreign technology providers.
In parallel, France plans a substantial investment of €655 million into its AI sector. This funding will support the development of sovereign AI tools, including a unified chatbot for all government services. A specific public health chatbot is also planned for the Ameli agency. Prime Minister Sebastien Lecornu highlighted the importance of France taking a leading role in the AI revolution and developing its own indigenous technologies.
However, challenges emerge regarding the capabilities of some French AI models. A study conducted by the Institute of the Estonian Language found that Mistral AI's models exhibit significant vulnerabilities, scoring below 40% in detecting Kremlin talking points. Some responses from these models referenced pro-Russian sources such as RT and Sputnik, raising concerns about their susceptibility to disinformation.
These developments coincide with international discussions on AI. Leaders at the G7 summit in Evian are slated to discuss the future of artificial intelligence. This agenda item follows the recent deactivation of two AI models developed by Anthropic, a U.S.-based AI safety and research company. The discussions are being highlighted on the Playbook Paris podcast.