In a recent appearance at the company’s inaugural AI conference in Paris, Arthur Mensch, the chief executive of French AI startup Mistral, clarified the firm’s stance on its technology’s use by defense customers amid ongoing ethical debates in the sector. Mistral, which has identified a significant opportunity in partnering with industrial giants like Airbus and BMW, also serves various defense operations, albeit its defense-related revenue constitutes only 10 to 15 percent of its total earnings.
Mensch emphasized, “Choices about deployment and usage are not our business.” He stated that the responsibility for how the technology is used lies with defense ministries, not with the companies providing the technology. Mistral’s software platform, which utilizes autonomous AI agents, can process and aggregate vast amounts of data, making it particularly useful for military operations, including tactical coordination on the battlefield.
The company recently secured a five-year partnership with Airbus, extending into its defense operations. Mistral’s smaller AI models could be integrated into defense systems like drones, enhancing their functions through autonomous capabilities. Mensch argued, “The defense ministry has considerably more legitimacy (to make decisions) than us,” distancing Mistral from what he characterized as “ideologues” who criticize technological applications in military contexts.
In contrast, other firms, such as Anthropic, have taken a firmer ethical stance regarding military applications. Anthropic, which promotes itself as the AI developer most committed to ethical concerns, has restricted the use of its systems by the U.S. government for purposes like mass surveillance or fully autonomous weaponry. This has led to legal disputes with the Pentagon.
Moreover, employees at Google’s DeepMind have voiced their discomfort with the military applications of their technology, further illustrating the growing unease among tech workers about the use of AI in warfare. Despite these broader concerns, Mensch acknowledged that Mistral has the discretion to choose its partners while ensuring “total sovereignty over weapons operation.”
Alongside its military applications, Mistral is working towards the development of Artificial General Intelligence (AGI), aiming for systems that may equal or exceed human intelligence across multiple disciplines. Mensch highlighted the necessity of implementing controls to prevent potential misuse of such advanced systems, stating, “We always deploy models and systems in environments where we can unplug them.”
Security remains a foundational priority for Mistral, established in 2023. The firm is also venturing into cybersecurity solutions tailored for companies, especially given that European firms—such as major banks—currently lack access to advanced models capable of uncovering and exploiting security vulnerabilities.
Mensch insisted that the evolving nature of cyber threats, particularly from adversaries equipped with AI, underscores the need for robust cybersecurity measures. He announced that Mistral’s cybersecurity product is set to be released later this year, aimed at fortifying defenses against sophisticated attacks.
